In his recent piece, “America Risks Blowing the AI Race. Here Are Seven Ideas to Get Back on Track,” Jim VandeHei is right about the stakes, right about the narrowing window, and right that America is improvising when it should be executing. But he misses one crucial fact: the playbook for winning this race was already written.

Five years ago, the National Security Commission on Artificial Intelligence — a bipartisan body created by Congress, chaired by Eric Schmidt, with commissioners including Safra Catz, Andy Jassy, and Eric Horvitz — delivered a 750-page blueprint anticipating nearly every problem he describes: the talent crunch, the workforce dislocation, the biosecurity risk, the need for an allied technology coalition. I served as the Commission’s executive director, and we didn’t just diagnose the challenge. We handed the President and Congress draft legislation and executive orders to meet it. The problem was never a failure of foresight. It was a failure of follow-through.

The tragedy isn’t that we failed to anticipate the future. It’s that we anticipated it, wrote it down, and then stopped short of implementing it. The good news, as VandeHei says, is that the window hasn’t closed. Here is how to turn his seven ideas into results using a playbook that already exists, and finish the job.

1. We’ve organized as a nation before. The space program shows how.

The moonshot is the right frame, but the deeper lesson comes from what preceded it. After Sputnik, America didn’t just build rockets. It built institutions: the National Security Council apparatus matured, NASA was created, DARPA was born, and the government forged enduring partnerships with industry and universities. We invested in science at scale. NSCAI’s version of that investment was concrete: double federal non-defense AI R&D annually to reach $32 billion per year by 2026. That number was not aspirational padding — it was calibrated to what winning requires. Today federal non-defense AI R&D sits at a small fraction of that figure while China funds AI as a wartime priority. Private capital is surging, but venture dollars don’t fund the foundational research, testbeds, and national research infrastructure that produced the internet, GPS, and AI itself. A Sputnik response means public investment on a Sputnik scale.

2. The coordinating body has a name: the Technology Competitiveness Council.

VandeHei calls for a working group with actual authority. NSCAI proposed exactly that: a Technology Competitiveness Council at the White House, chaired by the Vice President, with the standing and staff to drive a national technology strategy across agencies — the way the National Security Council drives foreign policy. Ad hoc task forces and czars evaporate with each news cycle. A council anchored in statute, with a mandate to map risks and build playbooks before crises hit, is how you institutionalize anticipation instead of improvising after the fact.

3. A National Defense Education Act 2.0.

The workforce displacement debate is unproductive because we are arguing instead of building. After Sputnik, Congress passed the National Defense Education Act and transformed American science education within a generation. NSCAI called for a second NDEA — a national investment in digital talent spanning K-12 STEM, scholarships-for-service, a U.S. Digital Service Academy, and a National Digital Reserve Corps. The Federal Government should consider how to implement NSCAI’s recommendation for a National Education Act 2.0 now, before unemployment numbers force a panicked version of it later.

4. Retraining should be a national system, not a corporate pilot.

A real-time jobs app is a good idea — and it needs to plug into something. NSCAI recommended treating workforce transition as a national security function: continuously mapping where AI is creating and destroying demand, funding retraining pipelines through community colleges and apprenticeships, and using government’s convening power to match displaced workers to the jobs the AI buildout itself is creating — data centers, energy, chips, technicians. Scattered corporate retraining pledges are welcome. A national system, informed by real-time data from the AI companies, is what scales.

5. Make AI literacy mandatory.

Preparing businesses, colleges, and families cannot be voluntary and episodic. NSCAI’s talent recommendations rested on a simple premise: AI literacy is a baseline civic skill in this era, the way basic scientific literacy became one after Sputnik. Every high school and college should teach it; every federal worker should be trained in it; and teachers should be equipped to use it rather than left to police it. A country that sends its students and workforce into the AI era blind has chosen to lose.

6. Build the institution that takes catastrophic risk seriously.

VandeHei is right that the dangers are knowable — and knowable danger demands institutional response, not episodic alarm. We have a model. When chemical and biological weapons emerged as threats, the Department of Defense built dedicated organizations to understand them, red-team them, and prepare defenses. We should do the same for AI: a standing office within DoD charged with mapping what state and non-state actors could do with frontier capabilities — cyber, bio, autonomy — and building response infrastructure before the capability arrives. NSCAI warned about AI-enabled threats from adversaries exploiting open societies. Preparation is a bureaucratic choice. We should make it.

7. Win the coalition by controlling the full tech stack.

The global coalition VandeHei proposes only works if there is something to join. That means organizing allies around the entire technology stack — chips, tools, models, cloud, energy, standards — so that the world’s AI ecosystem runs on American and allied technology rather than drifting toward Chinese alternatives. NSCAI called for exactly this: coordinated export controls with allies, a protected semiconductor supply chain, and an alliance framework for AI innovation among democracies. The offer to partners is enticing precisely because it is complete: join a stack that is more capable, more trusted, and more open than Beijing’s. Half-measures — controlling chips while ceding infrastructure deals in the developing world — leave the door open.

The window is still open

VandeHei writes that we knew what was coming two years ago. In truth, we knew more that eight years ago – that’s why Congress created NSCAI. And since then, Congress and two administrations adopted real pieces of the NSCAI agenda, from CHIPS to AI offices across government. But we executed perhaps a third of the playbook. China, meanwhile, executed its plan by decree. Democracies can’t and shouldn’t work that way. What we can do is what we did after Sputnik: decide, invest, build institutions, educate a generation, and rally allies.

The ideas are not missing. The blueprint is not missing. What has been missing is the will to execute. Shame on us if, having written the plan twice, we fail to run it.

Most rulemakings don’t make headlines. This one shouldn’t slip by unnoticed.

The Office of Management and Budget has proposed the most sweeping revision in years to the regulations governing federal grants and research awards. Buried in the administrative language are provisions that would fundamentally change how America funds science: senior political appointees would review every discretionary award, reviewers would be directed to weigh presidential policy priorities, peer review would be explicitly downgraded to “advisory,” and agencies could terminate active research awards simply because priorities changed after the work began.

Let me be clear about where I stand. Accountability for taxpayer dollars is not optional. Executive Branch officials are responsible for how federal funds are spent, and they should be. SCSP supports OMB’s goals of transparency, research integrity, and stewardship. Political accountability and scientific expertise are not enemies.

But this rule, applied without clear limits, risks damaging something we cannot afford to damage right now: the innovation engine that underpins American power.

We are in a competition, and the margins are thin.

At SCSP, we spend our days measuring where the United States actually stands against China in the technologies that will decide this century. The picture should concern everyone. Our 2026 Tech Competition Scorecard found that China holds a decisive overall lead in robotics for advanced manufacturing. In quantum, the United States is only slightly ahead — and that lead is eroding under Beijing’s coordinated, state-backed strategy. Meanwhile, the federally funded share of national R&D dropped by nearly one-third between 2010 and 2019, and federal AI R&D spending remains far below the $32 billion annual level the National Security Commission on Artificial Intelligence recommended.

That is the backdrop against which this rule should be judged. We already have a funding gap. We should not compound it with a confidence gap.

Science is not procurement.

The proposed rule borrows a termination-for-convenience model from government contracting. But a research grant is not a purchase order. It supports laboratory teams, doctoral researchers, unique data, custom equipment, and multiyear experiments. Terminate that work midstream and you don’t just stop a project—you destroy value that reimbursement cannot recover: people, data, momentum, and the private co-investment that rode alongside the federal dollar.

And that’s the piece Washington too often misses. Federal research policy doesn’t just affect universities. Companies, investors, national labs, and startups all make decisions based on whether federal programs are stable and merit-driven. Venture capital flows toward fields that federal research has validated and de-risked. When award decisions look political rather than technical, or when a grant can vanish because priorities shifted, private partners hedge, talent looks elsewhere, and the whole geometry of innovation weakens. Our competitors, operating under long-term national strategies, do not have this problem.

What we told OMB.

Our comment makes four core recommendations:

1. Keep merit at the center. Scientific and technical merit, evaluated by qualified experts, should remain the primary basis for selecting research proposals. Senior appointees have a legitimate oversight role—compliance, security, program fit—but they should not substitute their judgment for expert evaluation of scientific quality, and any override should be documented in writing against published criteria.

2. Make awards durable. Competitively awarded research grants should not be terminable simply because policy priorities changed after the fact. Reserve termination for real problems—legal violations, security concerns, performance failures—and give recipients a chance to respond and wind down responsibly.

3. Account for the private sector. Before changing how research awards work, agencies should ask whether the change will deter co-investment, interrupt commercialization pathways, or push globally mobile talent toward our competitors.

4. Assess the competitive impact. OMB’s own analysis of this rule counts only paperwork costs and calls them “modest.” It never asks the bigger question: what does this do to America’s position in the technology competition? Forgone discoveries and deterred capital are real national-security costs, even if they never show up as a line item. Major changes to federal research policy should come with a technology-competitiveness impact assessment.

We also urged OMB to slow down. An October 1 effective date—roughly 90 days after comments close—would impose new selection and termination frameworks on fiscal year 2027 awards before agencies have built the procedures to implement them well.

The real question.

The debate here is not whether the government should control taxpayer funds. It should. The question is whether the rules governing that control reinforce or undermine the ecosystem that made America the world’s leading scientific and technological power: technical excellence, merit-based evaluation, long-term inquiry, and productive connections among government, academia, industry, and capital.

China is aligning every instrument of national power around technology leadership. We should be strengthening our innovation base, not introducing uncertainty into it. OMB can achieve real accountability and preserve what works—but only if it revises this rule before finalizing it.

Our full comment is available here.

Explore How AI is Reshaping the Scientific Pipeline

Scientific breakthrough has historically been limited by the speed of human trial and error. Today, AI is fundamentally changing that equation by accelerating the R&D pipeline and running autonomous scientific agents to design and refine experiments.

To explore this shift, SCSP is hosting the AI+ Discovery Summit on July 21, 2026, in Washington, D.C.

The event is tailored for researchers, policymakers, and builders working at the intersection of deep tech and national competitiveness. Key themes include:

• Machine Discovery: Models designed to accelerate breakthroughs.

• Scientific Agents: Autonomous systems that can design, run, and refine search.

• Science Infrastructure: Managing the massive compute, data, and talent demands of AI-driven research.

Speakers include directors and leads from Los Alamos and Oak Ridge National Laboratories, Commonwealth Fusion Systems, Convergent Research, and more. Space is limited, so register now before it’s too late!

Register Now

This Week on the President’s Tech Brief

AI Sovereignty: Is AI the New Oil?: Fresh off Palantir's new partnership with NVIDIA, lead AI architect Chad Wahlquist joins to explain why he sees AI models as a commodity — and why he argues America should treat owning its AI infrastructure the way it treats protecting its oil reserves. He breaks down the "token maxing" trap that's leaving enterprises with unexpectedly large bills, why most workflows don't need a frontier model at all, and the analogy that reframes how companies think about the AI they're paying for.

Watch Here

NATO in Türkiye: A Summit Putin Won't Like: Former U.S. Ambassador to NATO Kurt Volker joins live to recap the summit in Türkiye — and to push back on the headlines about friction tearing the alliance apart. He lays out what actually came out of the meeting on European defense spending, missiles for Ukraine, and Zelensky's reception, and gives his read on how close this all pushes Putin toward a ceasefire.

Watch Here

Iran: The Ceasefire Roller Coaster: Iran struck three vessels near the Strait of Hormuz, the U.S. answered with roughly 170 strikes, and U.S. Treasury cut off the waivers keeping Iran's oil revenue flowing. Chip Usher and Ylber Bajraktari break down the one clause in the June agreement, Article V, that has become the whole ballgame — and why Chip says Iran is "in it to win it." There's also a ticking clock hiding in America's strategic petroleum reserve.

Watch Here

China Watch: Beijing's Own "Mythos Moment": David Lin and Austin Huang cover three stories that all trace back to one tension: Beijing wants to control the chips and AI models it depends on, even as its own companies race to break free of that control. DeepSeek is designing its own chip, the door to NVIDIA chip sales in China is opening slightly, and there's a reported plan that could change who gets access to China's most powerful models.

Watch Here

AI in the Office: The Self-Organizing Research Paper: This week, Research Assistant Mason Tsai shows how he used Claude to build an interactive app that keeps his white paper organized — mapping every source to the right section, flagging where his argument needs more support, and catching sources that overlap. Watch is below:

We're live every Friday at 11 AM ET on YouTube. Subscribe so you never miss an episode — and send us your questions, comments, and feedback. Consider yourself briefed.

The Catalyst & The Historical Legacy

Landlocked, surrounded by historic adversaries, and long labeled a mere post-Soviet nation subject to Russian influence, lies Armenia. Bridging Europe and Asia, Armenia is a nation of 3 million with a diaspora of 10 million people. One must wonder what connection a small nation like Armenia has to draw major investments backed by American global tech giants. In June 2025, Firebird.ai, a U.S.-based artificial intelligence cloud and infrastructure enterprise co-founded by two Armenian-Americans, launched an initial $500 million investment in Armenia. Since February 2026, this project has grown to an additional $4 billion through a strategic partnership with Nvidia and has received critical export approval from the U.S. Government. The catalyst for this historic expansion relies on Armenia’s strategic integration of its global diaspora network, bridging domestic technical capacity with Western capital markets. Outside of its diaspora, Armenia attracts Western tech and defense companies by offering highly skilled, low-cost talent rooted in a strong STEM tradition, alongside generous startup tax breaks.

From an investment standpoint, Firebird.ai anchors the corporate initiative, aligning diaspora-led commercial ventures with national strategic goals, bringing Western technological resources into the region. Serving as the conductor was a senior official at Nvidia, another Armenian-American, who brought together the Nvidia ecosystem, Firebird.ai, and the Armenian government to implement their shared vision of high-tech integration between the United States and Armenia. The vision from American investors was fundamentally rooted in the institutional capacity of Armenia’s tech sector and its rigorous historical precedent in mathematics and sciences.

In fact, the nation carries an unforgettable legacy; during the late 20th century, Armenia operated as the ‘Silicon Valley of the Soviet Union,’ designing an estimated one-third of all Soviet mainframe and military electronics. Central to this hub was the Mergelyan Institute, which pioneered early Soviet computing by developing the Nairi and Razdan mainframe series – massive, high-powered central computers built to handle complex data processing for the state’s military and scientific sectors. Driven by a central Soviet economic mandate that designated the republic for high-tech research and development, Armenia cultivated the highest number of scientists and engineers per capita of any Soviet state. This intentional institutional specialization provided a technical foundation that allowed the country’s tech sector to endure long after the Soviet collapse.

State Modernization, Tech, & Talent

Armenia’s commitment to technology development is also reflected at the highest levels of government today. Armenia maintains a dedicated, cabinet-level Ministry of High-Tech Industry (MHTI), founded in May 2019 by Prime Minister Nikol Pashinyan by restructuring the former Ministry of Transport, Communication, and Information technologies into a specialized high-tech portfolio. Instead of just regulating the market, this new ministry immediately focused on building international partnerships. In 2022, Nvidia opened its office in Armenia. Led by local tech leaders, this corporate presence allowed Armenian engineers to work directly on complex global projects and proved the country’s technical capabilities early on. That same year, the outbreak of the Russo-Ukrainian War triggered a major influx of technical talent. Enabled by Armenia’s existing 180-day visa-free entry regime, more than 100,000 software engineers and other skilled professionals relocated from Russia to Armenia. This reinforced the country’s historic position as the technology center of the Soviet Union and accelerated the development of its modern innovation ecosystem. These new residents and over 1,200 newly registered IT firms immediately injected massive capital into the country, driving a historic 12.6% surge in real GDP growth in 2022, the nation’s fastest economic expansion in 15 years. To cement this wave of talent, the Ministry spent 2023 and 2024 mapping out a long-term industrial strategy, while local skills expanded through initiatives like grassroots training at the Armenian Code Academy.

Ultimately, it was these years of handling crises and mapping out policies that gave the government its confidence to launch its next major tech push. The MHTI has gone to great lengths to codify this progress through targeted initiatives, including the January 1, 2025, implementation of the Law “On State Support for the High-Tech Sector,” which by the end of the year provided 8.76 billion drams ($23.7 million USD) in direct financial aid. The Armenian government has also introduced programs like the 60% income tax refunds designed to attract and retain technical talent. Furthermore, the Ministry has aggressively expanded its framework to explicitly focus on artificial intelligence infrastructure, shifting its focus from regulation to a more active role in international technology partnerships.

Taken together, these developments created the conditions for a deeper technology partnership between Armenia and the United States. Along with growing Western engagement and warming relations in the region, this environment helped pave the way for the August 2025 U.S.-Armenia AI and Semiconductor Innovation Partnership and the subsequent cooperation necessary to support the deployment of advanced AI computing infrastructure.

The administrative groundwork established years prior directly cleared the path for the country’s most ambitious infrastructure project to date. What began as a $500 million regional AI and technology data center announcement in November 2025 quickly shifted into an ambition to build one of the world’s most powerful AI supercomputing clusters, subsequently sizing the project to a total $4.5 billion investment specifically scaled to accommodate the massive deployment of 50,000 NVIDIA Blackwell GPUs by the end of 2026. This unique joint initiative between Firebird, Nvidia, Dell Technologies, and the Armenian government was finalized after receiving U.S. Government export licenses and regulatory approvals in May 2026.

Located in the small lakeside town of Hrazdan, the megaproject is designed as an on-the-grid facility that strategically protects local residents from power shortages or grid strain by connecting directly to the Hrazdan Power Station. This allows the project to utilize the region’s available diverse nuclear, hydro, and solar energy. What also makes this specific project notable is that, unlike traditional facilities that heavily consume local water supplies for cooling, the Hrazdan site introduces an eco-friendly closed-loop water system. This advanced setup continuously recycles its water supply for years, virtually eliminating the local environmental and water-waste concerns typically associated with massive data centers. This clean electricity powers the facility’s first wave of NVIDIA Blackwell GPUs, assembled in Dell server towers and delivered to the Hrazdan site in June 2026, marking the beginning of the project’s operational buildout.

The project also has strategic significance for the United States. American businesses have exclusively reserved 80% of this massive computing capacity for purchase, leaving 20% for local Armenian enterprises, highlighting its strong alignment with U.S. commercial interests. Once all 50,000 advanced GPUs are deployed later this year, this immense hardware capacity could catapult Armenia into the top five countries globally for total number of graphics processing units. This scale would position Armenia as a significant regional computational node, a standing that is expected to strengthen with further expansion plans that look to scale the compute capacity in Armenia beyond 100,000 NVIDIA Blackwell and incorporate next-generation architecture Vera Rubin GPUs by the end of 2027.

This computational capacity requires a highly skilled domestic workforce, and Armenia is systematically matching these hardware investments with forward-looking talent development. The country is putting this immense computing power to use immediately through a pioneering partnership with OpenAI’s Education for Countries initiative. Funded completely by Firebird.ai, this milestone delivers 50,000 advanced subscriptions of ChatGPT Edu and Codex straight to local students, educators, and researchers to support AI-adoption. Prior to that development, grassroots learning hubs like the world-renowned TUMO Center for Creative Technologies, founded in Yerevan, have played a leading role in technology and design education for students. TUMO recently won first place and earned the prestigious WISE Prize for Education for its new AI learning platform, an innovation heavily supported by global tech leaders. Although Armenian in origin, TUMO has grown into an international network with centers spanning Europe, North and South America, and Asia, illustrating how Armenia is exporting its educational innovation alongside its technological ambitions.

The Geopolitical Pivot to the West

This combination of domestic human capital and localized raw computing power signals a geopolitical shift, pulling Armenia out of Russia’s shadow and placing it in a more secure position for American and European businesses. The recent reelection of Prime Minister Nikol Pashinyan signals a continued momentum towards the West and away from old dependencies on Russia. Since Pashinyan first came to power in 2018, there has been a well-noticed shift in public and political attitudes towards welcoming American businesses into the country, where 42% of Armenians name the United States as one of their country’s most important political partners.

Historically under the Kremlin’s influence, Armenia’s emergence as a trusted technology partner has reinforced broader efforts to deepen ties with Washington. For U.S. strategic planners, this shift offers a calculated opportunity to secure critical supply chains and data infrastructure before adversarial actors can dominate the regional technology landscape.

U.S. Secretary of State Marco Rubio cemented this growing diplomatic trust during his historic working visit to Yerevan on May 26, 2026, when he met with Armenian Foreign Minister Ararat Mirzoyan. His trip marked the first time a U.S. Secretary of State had visited Armenia in ten years. During this milestone meeting, the two countries formalized a Comprehensive Strategic Partnership, reflecting the growing alignment between Armenia and the United States. The two nations also finalized the Trump Route for International Peace and Prosperity (TRIPP) framework, a strategic trade and transit corridor designed to bypass adversarial trade routes and secure Western supply chains through the South Caucasus. This growing economic partnership allowed the two nations to look beyond technology and sign a vital Critical Minerals Framework Agreement on the supply and processing of rare earths, directly strengthening allied supply chain resilience. The United States has an urgent strategic need to diversify its raw material supply chains away from monopolistic sources, meaning Armenia’s reserves of copper, molybdenum, and zinc could provide American technology and defense manufacturers a critical advantage.

This unprecedented alignment of hardware and human capital signals a newfound geopolitical transformation, systematically shifting Armenia away from traditional Russian influence and anchoring it firmly within the sphere of U.S. economic interests. By anchoring a 4.5 billion dollar computing infrastructure within its borders and optimizing it through elite educational programs with TUMO, OpenAI, and critical mineral agreements, Armenia may be demonstrating a new model for national sovereignty. This public-private infrastructure alignment introduces a complementary framework to traditional bilateral aid agreements, anchoring regional stability in supply chain integration and digital sovereignty. For the United States, this is a significant strategic opportunity to secure critical raw materials and digital ecosystem alignment through private enterprise and capital deployment. If current projections materialize, Armenia’s emergence as a major AI infrastructure hub could elevate a democratic partner into the ranks of the world’s leading centers for advanced computing capacity. This strategic position is sustained by the institutional strength of the nation, state-led resource optimization, and the collaborative capacity of its global network.

In Tokyo, SCSP President Ylli Bajraktari and VP for Global Affairs Joe Wang sat down with Taro KONO — former Foreign Minister, Defense Minister, and Digital Minister of Japan, and one of the most influential figures in the LDP — for a wide-ranging conversation on Japan’s place in a rapidly shifting world.

The conversation covered four major themes shaping Japan’s strategic future:

The Geopolitical Landscape: Kono traces how China’s rise changed everything — from its WTO accession to its island-building in the South China Sea — and why Japan will need more than just the U.S.-Japan security treaty alone to guarantee stability in East Asia.

Trump, Beijing, and the Alliance: Kono offers a candid assessment of Trump’s Beijing visit, which may have embolded Xi to act more aggressively against Japan and others in the Indo-Pacific, and why coordinated allied policy vis-à-vis China is urgently needed.

Japan’s Defense Transformation: From increasing defense spending and building a robust domestic defense industry, Kono lays out the path Japan must take to shoulder more of its own security burden.

AI, Autonomy, and Japan’s Digital Future: Kono reflects on his time as Digital Minister, Japan’s demographic crisis driving AI adoption in government, the rise of drone warfare, and why Japan sees a “quad plus” technology alliance as the realistic path forward.

Apple

Spotify

🔬 Mapping the Frontier of AI-Powered Science: Join SCSP on July 21st!

AI isn’t just changing how we work—it is fundamentally reshaping the rules of scientific discovery. To map out this new landscape, SCSP is excited to host the AI+ Discovery Summit next month.

📅 When: July 21st |📍 Where: Washington, D.C.

This gathering will bring together top minds from across industry, science, and government, including Dr. Georgia Tourassi (Oak Ridge National Laboratory), Dr. Brandon Sorbom (Commonwealth Fusion Systems), Anastasia Gamick (Convergent Research), Dr. Olaf Groth (Cambrian Futures), and more. From exploring next-generation breakthroughs to diving into the Department of Energy’s Genesis Mission, we will outline the strategic blueprint needed to secure America’s tech leadership.

Register for the AI+ Discovery Summit

Disinformation has always shadowed elections, but artificial intelligence now lets bad actors produce it at a scale and speed never seen before. With the midterms approaching and government election-security agencies reduced, the question grows urgent: Can the same technology fueling the threat also be used to defend against it?

Mark Murray joins NatSec Tech to share his perspective on AI, disinformation, and election integrity. Mark is Editorial Director at OpenAI and previously served as senior political editor at NBC News. Drawing on decades covering elections and his work at the frontier of AI, he explains how AI can empower both bad actors and good ones, how tools like digital watermarking and verification aim to combat deepfakes, and why he's confident the midterms can still be free and fair.

Listen to the Episode

🔬 Mapping the Frontier of AI-Powered Science: Join SCSP on July 21st!

AI isn’t just changing how we work—it is fundamentally reshaping the rules of scientific discovery. To map out this new landscape, SCSP is excited to host the AI+ Discovery Summit next month.

📅 When: July 21st |📍 Where: Washington, D.C.

This gathering will bring together top minds from across industry, science, and government, including Dr. Georgia Tourassi (Oak Ridge National Laboratory), Dr. Brandon Sorbom (Commonwealth Fusion Systems), Anastasia Gamick (Convergent Research), Dr. Olaf Groth (Cambrian Futures), and more. From exploring next-generation breakthroughs to diving into the Department of Energy’s Genesis Mission, we will outline the strategic blueprint needed to secure America’s tech leadership.

Register for the AI+ Discovery Summit

It's Taiwan's most famous export in a cup, but behind every chewy tapioca pearl is a global business story of culture, competition, and surprising technology. In this episode, host Channing Lee talks about all things boba with Greg Chen, CEO of International Business and Chief Investment Officer for the group behind Sharetea and Unique Green (UG) Tea, two of Taiwan's biggest bubble tea brands. They cover where boba came from, how the drink became a worldwide phenomenon and a symbol of Taiwanese culture, why Chinese chains flooding the market pose a growing competitive threat, and how AI-powered machines are helping stores leap from 100 cups a day to thousands.

Apple

Spotify

🔬 Mapping the Frontier of AI-Powered Science: Join SCSP on July 21st!

AI isn’t just changing how we work—it is fundamentally reshaping the rules of scientific discovery. To map out this new landscape, SCSP is excited to host the AI+ Discovery Summit next month.

📅 When: July 21st |📍 Where: Washington, D.C.

This gathering will bring together top minds from across industry, science, and government, including Dr. Georgia Tourassi (Oak Ridge National Laboratory), Dr. Brandon Sorbom (Commonwealth Fusion Systems), Anastasia Gamick (Convergent Research), Dr. Olaf Groth (Cambrian Futures), and more. From exploring next-generation breakthroughs to diving into the Department of Energy’s Genesis Mission, we will outline the strategic blueprint needed to secure America’s tech leadership.

Register for the AI+ Discovery Summit

Futurist Ari Wallach joins Intel at the Edge to discuss why the U.S. government canceled its long-range forecasting program — and what it costs us. From scenario planning and megatrends to AI, loneliness, and narrative warfare, Wallach makes the case for embedding foresight into how America thinks and acts.

Wallach is the founder of Long Path Labs and the former host of the PBS television series A Brief History of the Future. He has advised the Department of State, the United Nations Refugee Agency, and consulted on strategic foresight — the discipline of mapping not just what is likely to happen, but what could happen, and how to recognize the early signals that tell you which future is taking shape.

Apple Podcasts

Spotify

🔬 Mapping the Frontier of AI-Powered Science: Join SCSP on July 21st!

AI isn’t just changing how we work—it is fundamentally reshaping the rules of scientific discovery. To map out this new landscape, SCSP is excited to host the AI+ Discovery Summit next month.

📅 When: July 21st |📍 Where: Washington, D.C.

This gathering will bring together top minds from across industry, science, and government, including Dr. Georgia Tourassi (Oak Ridge National Laboratory), Dr. Brandon Sorbom (Commonwealth Fusion Systems), Anastasia Gamick (Convergent Research), Dr. Olaf Groth (Cambrian Futures), and more. From exploring next-generation breakthroughs to diving into the Department of Energy’s Genesis Mission, we will outline the strategic blueprint needed to secure America’s tech leadership.

Register for the AI+ Discovery Summit

Hello, and welcome to a special edition of the SCSP newsletter.

Two hundred and fifty years ago, a small group of people made an audacious bet: that ordinary citizens, governing themselves under a set of ideas rather than a king, could build something stronger and more free than anything that came before. It was, by the standards of its day, a wildly improbable startup. It worked.

We tend to forget that America was a competitive project from the very first sentence. The Declaration was not a description of the world as it was; it was a wager on the world as it could be. Every generation since has been handed the same assignment — to take an unfinished idea and make it more true, more capable, and more secure than they found it.

That is the right way to read this Fourth of July. We sit at an inflection point as consequential as any in our history, with artificial intelligence and a wave of emerging technologies reshaping our economy, our security, and our society.

The pessimists will tell you the American era is ending, and they are not making the argument from thin air. Last year, when the Chinese lab DeepSeek released a competitive AI model said to be trained at a fraction of the expected cost, it erased roughly $600 billion from a single American chipmaker’s value in one day and launched a thousand headlines declaring that the United States had already lost the AI race. Add it up, the declinists say — China’s lead in patents filed and STEM doctorates granted, its dominance in shipbuilding, batteries, and the industrial base, and a manufacturing sector that continues to expand on the back of AI hardware and other high-tech exports. The trajectory looks unmistakable.

The second indictment comes from within. A national debt north of $39 trillion, a political polarization so deep that the two halves of the country can scarcely agree on a shared set of facts let alone a shared future, politics that too often mistake noise for governance, and eroding trust in our own institutions are, to the pessimist, not passing storms but the symptoms of a great power that has lost the plot. A house this divided, the argument goes, cannot summon the unity of purpose that great competitions demand. Every empire eventually declines, the pessimist concludes, and ours is simply next in line.

We have heard this song before, about Sputnik, about Japan’s ascendancy in the 1980s, about every moment when the obituary writers sharpened their pencils for America. And we still see it differently. SCSP has named the antidote, and built our entire agenda around it: strategic confidence. We have written plainly that disillusionment is real and that pessimism can paralyze action. But the answer to that is renewed purpose, not surrender. As H.R. McMaster has long argued, resignation and defeatism are choices and so are confidence and agency. We are not passengers being carried along by forces beyond our control. We are, as we have always been, the authors of what comes next.

So at 250, let’s do something the moment calls for: take honest stock of how much is going right.

Congress saw this coming. When it stood up the National Security Commission on Artificial Intelligence, the bipartisan effort that many of us went on to carry forward into SCSP, it asked a hard question: was America ready for the world that AI was about to remake? The Commission’s answer was blunt. The United States was not yet “AI-ready,” it warned Congress and the President, and it had only until 2025 to get there.

But the Commission was no doomsayer. Its 2021 Final Report paired that warning with a rallying cry that has only grown truer with time: “AI is going to reorganize the world. America must lead the charge.” It had said as much from the start; the very first of its 2019 interim judgments held that “Global leadership in AI technology is a national security priority.” And it was clear-eyed about how this competition is won — that advantage flows to whoever moves first, that whoever turns AI breakthroughs into real-world applications soonest will set the terms for everyone else. In other words: the future does not simply happen to a nation. It is built, fielded, and won — or it is ceded.

That conviction ran through the integrated strategy the Commission laid out, and it has driven SCSP’s work ever since — across reports, commissions, and convenings aimed squarely at strengthening American competitiveness. Several years on, the picture is brighter than the doubters predicted. Here is the good news.

Reasons for Optimism

1. American innovation is alive and kicking. The frontier of artificial intelligence is being defined, overwhelmingly, by American companies, American capital, and American compute. The dynamism that built the railroad, the airplane, the microchip, and the internet did not retire, it moved into a new arena. SCSP has a name for this asset: innovation power — the decisive edge that flows from a free people’s ability to think, question, experiment, and build together. It is the one form of power our authoritarian rivals cannot simply copy by decree, and it remains the envy of the world.

2. Our universities still produce the world’s best talent. The American research university remains the single greatest magnet for ambitious, brilliant people on the planet. They come here to study, to invent, to start companies, and — when we let them — to stay. Talent is the holy grail of this competition, and the United States starts with the deepest bench in history.

3. Government is adopting AI like never before. What was once a slow-moving bureaucracy is now moving with real urgency. The idea we championed at NSCAI — that the White House needs to organize itself to win the technology competition — has moved from recommendation to reality, and SCSP’s ideas have helped shape the national AI agenda, including the federal AI action plan. The institutions of government are learning, at last, that adoption is not optional.

4. The military has internalized that the time to adopt is now. When we wrote that the Department of Defense and Intelligence Community had to be AI-ready by 2025, it was an aspiration that many called impossible. Today, the warfighting enterprise is reorganizing around it — building digital infrastructure, rethinking acquisition, and treating AI readiness as a measure of military readiness. The cultural shift from “someday” to “now” is the single most important development in American defense in a generation.

Want to see this momentum in action? We recently brought together the brightest minds in tech, government, and defense to showcase how America is actively building the future. Watch our recap of the AI+ Expo 2026 below:

The Sources of Our Strength We Must Protect

Optimism is not complacency. The same self-renewing instinct that built this country requires us to defend the foundations it stands on. Those foundations are enduring, but they are not self-sustaining. They must be renewed by every generation.

Immigration is America’s original force multiplier. I don’t write that as an abstraction; I write it as an immigrant myself. I am one of the millions who left the familiar behind to bet on the American idea and found here the room to build a life and a purpose. The people who make that choice are not a cost to be managed. They are the reason the experiment keeps outperforming. A nation that welcomes the world’s most determined talent, and gives it room to build, wins the long game by default. A nation that makes it harder for the world’s brightest scientists, engineers, entrepreneurs, and students to come here is not protecting its competitive edge — it is exporting it. Protecting and modernizing the pathways that bring that talent here is a competitiveness imperative, not a footnote.

Science is the engine that drives the country forward. Nearly every advantage we are celebrating this Fourth of July traces back to decades of patient public and private investment in fundamental research. We must ensure our R&D foundation continually strengthens, and is not weakened. The next set of breakthroughs — in AI, in biotechnology, in fusion energy, where SCSP’s own commission has charted a path to American leadership — will come from the same source. A country that stops investing in discovery stops shaping the future and starts inheriting someone else’s.

Capable public institutions turn innovation into national strength. America’s greatest breakthroughs have never come from the private sector or government acting alone, but from the partnership between them. From DARPA and NASA to the national laboratories and the defense innovation ecosystem, our public institutions have repeatedly transformed breakthrough science into strategic advantage. That requires attracting, empowering, and retaining exceptional public servants — from scientists and engineers to diplomats, acquisition professionals, and military leaders. In a world defined by technological competition, government capacity is not bureaucracy; it is a national asset.

Our allies are our asymmetric advantage. This is the strategic fact our competitors cannot replicate. The authoritarian powers seeking to revise the international order — China, Russia, Iran, and North Korea — have transactional partnerships of convenience. We have a coalition of free and capable nations bound by shared values and interoperable strength. Reinvigorating and scaling that network — across defense, technology standards, supply chains, and talent — is how a coalition of democracies out-innovates and out-competes any rival bloc. We should treat our alliances as the crown jewels they are.

The Work Ahead

Here is the most hopeful thing about all of it: none of this is automatic. The future of the American era will not be settled by some impersonal tide of history — it will be settled by what we choose to do, starting now, and we have done harder things than this before.

We begin by naming our gaps honestly and then closing them, because confidence and candor have always been partners rather than opposites. The slow seams in government, the thin spots in our talent pipelines and our industrial base — these are not verdicts handed down on us. They are a to-do list, and Americans have never been more themselves than when they roll up their sleeves over a to-do list. Every item on it is fixable, and fixing it is about as patriotic as work gets.

Then comes the part we should be most excited about: imagining a future unlike any before. The technologies converging around AI put within reach things our founders could only have called miracles — cures for diseases that have stalked humanity for millennia, discovery that compounds on itself, abundance that lifts people up rather than leaving them behind. We don’t have to stumble into that future and hope it turns out well. We get to design it on purpose, with American optimism and American values built into the foundation.

And we build it for everyone. This means we build it safely, especially for our children. A technology revolution worthy of this country develops and fields these tools responsibly, in line with the democratic values that make the whole project worth defending, treating the wellbeing of the next generation not as an afterthought but as the entire point. Winning the competition would mean very little if we lost sight of the people we are competing for.

That, in the end, is the thing Americans have always done best — not merely guarding what already exists, but building what does not yet exist, and throwing the doors open so everyone can help build it. We have been at it for 250 years. The truth is, we are just getting warmed up.

When NSCAI delivered its final report, it adapted a line from Churchill: we were at the beginning of the beginning. That is even truer today, and it is exactly the right way to mark a 250th birthday. America is not a finished monument to be guarded. It is a living project, the most successful one in human history, and the next chapter is unwritten.

A contest for the future is unfolding — and by the end of this decade we will know whether the world is shaped by free and open societies or dictated by censorship and coercion. We intend to make sure it is the former. We will help write that chapter through our reports and commissions, our convenings like the AI+ Expo, the next generation of leaders we are developing through the International Strategy Forum, and a community of people who believe, as we do, that the future can be better than the past if we build it that way.

Our founder, Eric Schmidt — who chaired the NSCAI before launching SCSP — puts the whole endeavor in five plain words: “I want America to win.” So do we. And at 250, the tools to do exactly that are more firmly in our hands than the pessimists will ever admit.

Happy 250th, America. The best is still ahead — if we choose it.

— SCSP

Lethal Autonomy: When the Machines Fly Better Than the Pilots: The Honorable Frank Kendall, Former Secretary of the Air Force, joins to discuss his new book, Lethal Autonomy: The Future of Warfare — Whether We Like It or Not. He explains why the era of American conventional dominance is closing, why our legacy platforms are headed for obsolescence, and what he witnessed firsthand in an F-16 during AI-versus-human dogfight tests. His conclusion about who wins that fight — human or machine — is one the U.S. military isn't yet ready to hear. SCSP’s Justin Lynch also joins the conversation.

Watch Here

Iran: The Groundhog Day Conflict: It’s day 124 of the Iran Conflict and Ylber Bajraktari and Chip Usher break down the latest — Iranian strikes on tankers, U.S. airstrikes in southern Iran, an Israel-Lebanon framework that Hezbollah instantly rejected, and a Wall Street Journal report that the Pentagon is refreshing its military options. Chip explains why he now believes the June MOU was never designed to resolve the conflict — and why this "managed conflict" could grind on all the way to the midterms. Plus: why the oil markets are far more optimistic than the physical supply picture justifies.

Watch Here

China Watch: Are We Even Running the Same Race?: With U.S. frontier models offline for 18 days after the Fable shutdown, China released GLM 5.2 — and startups started saying out loud they might switch to Chinese open-weight models. David Lin and Channing Lee break down what that moment revealed about the competition, and Channing makes the case: the U.S. and China aren't racing toward the same finish line at all. One is chasing frontier capability; the other is chasing global adoption — and may already be winning the race that matters more. Plus: why Alibaba's distillation campaign may be the strongest case yet for enforcing the Pentagon's blacklist.

Watch Here

AI in the Office: The Analyst's Ensemble: This week, Senior Adviser and former CIA analyst Chip Usher shows how he wired together three tools — OpenRouter, Hermes, and Obsidian — into a personal, source-grounded research system that answers questions from a library he actually trusts. He walks through exactly how it works. Watch it below.

We're live every Friday at 11 AM ET on YouTube. Subscribe so you never miss an episode — and send us your questions, comments, and feedback. Consider yourself briefed.

Listen to Mr. Kitamura's Remarks

Theme 1: The G7 and the Fragmentation of the World Order

Mr. Kitamura’s read on the Évian G7 Summit was optimistic but unsentimental. Its real achievement wasn’t rhetorical unity—it was the recognition that democracies must finally fuse security, technology, energy, finance, supply chains, and values into a single framework. International order, he argued, is no longer defended by diplomacy or military deterrence alone, but by the full architecture of national power: intelligence, economy, and technology together. China, Russia, Iran, and North Korea are not a monolith, but they share one habit—hunting for the seams of open societies, betting that democracies are too slow and divided to respond.

His verdict: the free world is resilient, but not yet fast enough. The lesson he pressed hardest is that deterrence must be comprehensive—covering not just invasion but also coercion, not just cyberattack but also technology theft, and not just blockade but also supply-chain strangulation. His ambition for the G7 is for it to become less of a summit process and more of a strategic operating system—and the test is whether it can move from declarations to disciplined execution.

Theme 2: The Evolving U.S.-Japan Alliance

The alliance, Mr. Kitamura noted, has grown from a bilateral defense arrangement into a genuinely global strategic partnership—now central to the Indo-Pacific balance, the resilience of the global economy, and the defense of the rules-based order. Japan’s National Security Strategy, counterstrike capabilities, rising defense spending, and active cyber defense mark a historic shift: a sober acknowledgment that geography and economic interdependence are no longer enough. The deepest change, he said, is psychological—Japan is shedding the postwar habit of separating economy from security. In this age, economic security is national security.

But the work is unfinished. He called for faster, deeper intelligence sharing; a cyber posture that matches the threat; integrated defense industrial bases; and planning for simultaneous contingencies, since authoritarian states study our response cycles and may try to overload them. His call for the next stage is alliance integration—common threat assessments, technology-protection rules, supply-chain strategies, and operational planning across military and non-military domains, so that coercion fails before it begins.

Theme 3: Countering Beijing’s Revisionist Ambitions

Beijing’s objective, in Mr. Kitamura’s framing, goes well beyond territory: a regional and eventually global order in which the Chinese Communist Party’s system, leverage, standards, and military power define everyone else’s choices—making resistance costly, neutrality hard, and dependence normal. It advances most often without firing a shot, through coast guard pressure, maritime militia, economic retaliation, cyber intrusion, lawfare, and influence operations. The old bet that integration would moderate authoritarian behavior, he said, has failed.

His prescription rests on four pillars: military deterrence (joint command-and-control, missile defense, counterstrike, and the defense of the Southwest Islands, with peace in the Taiwan Strait indispensable to Japan’s security); economic deterrence (export controls, investment screening, and trusted supply chains prepared in advance); technology protection (guarding AI, quantum, semiconductors, and biotech without closing our societies); and coalition-building (with Australia, India, South Korea, the Philippines, Europe, and Southeast Asia). The goal, he stressed, is not to contain the Chinese people but to defend choice, sovereignty, and the rule of law—stability built on deterrence, not submission.

Theme 4: Weapons of Economic Coercion

On economic statecraft, Mr. Kitamura was blunt: we are doing more than before, but still not moving at the speed of the threat. Authoritarian states have practiced economic coercion for years with centralized direction and few constraints; democracies must answer without surrendering their openness. He flagged three exposures—critical infrastructure (ports, grids, cloud, undersea cables, and the dependence and hidden ownership behind them), predatory M&A (technology leaking out through investment, joint ventures, and licensing, which demands sharper and harmonized allied screening), and AI as a strategic asset (a general-purpose technology whose compute, chips, models, and talent must be protected).

But defense alone isn’t enough—economic security requires industrial strength: trusted production in semiconductors, batteries, critical minerals, pharmaceuticals, and next-generation energy. Here he sees a natural division of labor: Japan’s advanced manufacturing and materials science paired with America’s scale, capital, and AI ecosystems can form the core of an allied economic-security architecture. His central lesson: economic openness must be protected by strategic discipline—preserving openness where possible, restricting access where necessary, and building resilience everywhere.

Our thanks to Mr. Kitamura for his time, candor, and decades of service to the U.S.-Japan alliance and to the free world. Conversations like this one are a reminder that the competition ahead will not be won by technology or strategy in isolation—but by the democracies willing to integrate the two and move with the urgency the moment demands.

Read and Download the Report

China Is Winning Southeast Asia’s EV Race

In the late 1990s, Vietnamese entrepreneur Phạm Nhật Vượng made a fortune selling instant noodles in Ukraine. Two decades later, he set out to build something far more ambitious: a national electric vehicle (EV) champion. In 2017, his company VinFast broke ground on a massive automotive complex in northern Vietnam; just 21 months later, the factory was complete, and it produced 200,000 EVs in 2025.

VinFast’s rapid rise reflects a broader dynamic underway across Southeast Asia. Driven by attractive government incentives, affordable pricing, and rising fuel costs, Southeast Asia has quietly become the world’s fastest-growing EV sales market.

But this emerging landscape is being shaped by more than just domestic ambitions. Facing intense competition and overcapacity at home, automakers from the People’s Republic of China (PRC) are expanding aggressively across the region — investing heavily in factories, supply chains, and distribution networks.

In October 2025, SCSP published “Charging Ahead: How China is Driving Innovation to Dominate the Global Electric Vehicle Market,” highlighting how Chinese firms are surpassing global automakers. Today, we are releasing our second major analysis: “Charging Ahead: Southeast Asia: How China is Driving the World’s Fastest-Growing Electric Vehicle Market.” This report offers a unique case study of how the PRC is testing the durability of its industrial policy overseas by examining how governments and consumers in Southeast Asia are responding to the influx of Chinese EVs.

The State of Play: Explosive Growth

Southeast Asia’s EV adoption is rapidly outpacing many Western markets. Driven by an influx of affordable Chinese brands, BEV sales as a percentage of total vehicle sales have surged to over 50% in Singapore and 40% in Vietnam. By the end of Q1 2026, Thailand reached 33%, and Indonesia hit 16% — both comfortably outpacing adoption levels in the United States, which dropped to 5.5% over the same period.

While Vietnam and Thailand regularly alternate as the region’s largest absolute domestic markets, per capita data reveal that Singapore leads the region in adoption by a massive margin. However, the defining story across almost every country in the region is the dominance of Chinese parent companies, which collectively commanded 55% of the regional EV market share in 2025. Aside from VinFast’s strong domestic sales in Vietnam, no foreign automakers — American, German, or Japanese — come close to Chinese sales volumes.

A head-to-head comparison of U.S. and Chinese BEV sales across the region illustrates an even starker contrast, confirming that Chinese companies outsell U.S. ones in every Southeast Asian market.

A Tale of Three Countries

Our report dives into the region’s three largest EV markets — Vietnam, Thailand, and Indonesia — revealing three distinct competitive strategies:

• Vietnam, The Homegrown Monopolist: Vietnam boasts a unique advantage thanks to VinFast, which has maintained a staggering 99% domestic market share for the last five years. This success is heavily insulated by aggressive government mandates — including a planned ban on fossil-fuel vehicles by 2040 and local restrictions starting in Hanoi as early as July 2026. However, VinFast remains hyper-dependent on its home turf; 93% of its 2025 sales occurred domestically. To capture this market, Chinese giants like Chery and BYD are pivoting from pure exports to building local left-hand-drive factories and component plants in Vietnam.

• Thailand, The Chinese-Dominated EV Manufacturer: Thailand has leveraged its historic internal combustion engine (ICE) manufacturing ecosystem to position itself as the region’s premier EV manufacturing hub. Through aggressive “EV 3.0“ and “EV 3.5“ policies, Bangkok forced foreign automakers to match imported vehicles with local production quotas. Chinese firms enthusiastically accepted the challenge: BYD, GAC Aion, Changan, and Chery have all opened massive greenfield manufacturing plants in Thailand’s Eastern Seaboard. Consequently, Chinese brands command roughly 90% of Thailand’s BEV sales, systematically hollowing out the Japanese legacy automakers (including closures by Suzuki, Subaru, Honda, and Nissan) that historically dominated the country.

• Indonesia, The Resource-Rich Assembler: Indonesia represents a massive domestic market attempting to replicate Thailand’s manufacturing success, albeit a couple of years behind. Jakarta used zero-tariff import incentives to spur adoption, then transitioned to strict localization mandates (TKDN) requiring 40% local components. Chinese brands captured 95% of the market in Q1 2026, though structural hurdles remain. Indonesia possesses 60% of the world’s nickel reserves, but the global EV industry is rapidly pivoting from Nickel Manganese Cobalt (NMC) batteries to lithium iron phosphate (LFP) alternatives. This technological shift threatens Indonesia’s ability to stay relevant in the global EV supply chain.

The Shifting Frontier

Beyond these three largest EV markets, the trend of Chinese EV dominance is spreading across the rest of Southeast Asia. While pricing is less of a concern in Singapore, consumers are increasingly choosing to buy from Chinese companies, which accounted for 68% of BEV sales in Q1 of 2026. In Malaysia’s burgeoning market, the historic national car brand, Proton, captured 55% of the BEV market share in Q1 2026. However, Proton is a national champion with a twist: China’s Zhejiang Geely Group owns a 49.9% stake, effectively making Malaysia’s debut national EV Chinese-backed, both financially and technologically. Meanwhile, smaller markets like the Philippines, Laos, and Cambodia have eliminated import taxes, resulting in explosive tenfold sales growth since 2023 — almost exclusively driven by Chinese brands like BYD and Xpeng.

Strategic Implications for the United States

There is an old Chinese idiom: 反客为主 (fǎn kè wéi zhǔ) — meaning the guest becomes the host. In Southeast Asia’s EV ecosystem, China has done exactly that. What began as a venue to offload domestic overcapacity has solidified into a permanent economic and industrial footprint. Excluding VinFast’s majority sales in Vietnam, BYD is the region’s leading BEV company.

This creates an enduring strategic challenge. Just as Huawei’s integration into global 5G networks created long-term technical dependencies, the region’s deep reliance on Chinese EV infrastructure — from digital vehicle ecosystems to localized charging grids — may grant Beijing future political leverage in what it calls its “backyard.”

Furthermore, this shift marks a profound transition in the PRC’s global tech standing: China has officially evolved from a technology recipient to a primary source of overseas technology transfer.

For the United States, the current landscape is a stark wake-up call. American companies barely register in regional sales, capturing only 3.9% of the BEV market in 2025. While entering Southeast Asia may present lower near-term profit margins, local demand is undeniable. Bolstering U.S. regional engagement is critical to countering a deepening reliance on Chinese EV infrastructure.

Because American firms cannot easily compete on price with Chinese brands, Washington and U.S. automakers must change the parameters of the game. This means aggressively emphasizing safety, data security, and vehicle privacy, or focusing heavily on the emerging autonomous vehicle (AV) market as a fresh, mature entry point.

For now, however, Chinese companies are charging Southeast Asia.

Read the Report

If you only have two minutes today, watch these:

Iran: A Deal That Manages the Conflict — But Doesn't End It?: It's day nine of the U.S.-Iran memorandum of understanding, and the early signs aren't encouraging: ships are moving through the Strait of Hormuz again, but Iran struck a Singapore-flagged vessel just yesterday. Chip Usher breaks down why he believes this agreement was never designed to resolve the conflict — only to manage it — and the weaknesses buried in the text, from the missing ballistic-missile file to a clause he calls "a slap in the face" to the Iranian people. With 51 days left to reach a nuclear deal, he explains why this one is tenuous at best.

Watch Here

Pax Silica: Can the West Finally Break China's Grip?: A major coalition of nations — now including the EU, the Netherlands, Germany, and several Western Hemisphere countries — has signed onto a unified framework to secure critical minerals and challenge Beijing's monopoly. Ylli and Martijn break down what the summit actually achieved, why this round looks different from past efforts, and the one thing Ylli worries could still sink it. His warning: if we're having this same conversation in 2028, we will have already lost to China.

Watch Here

Americans Love AI — and Fear It. Why?: BJ Martino of The Tarrance Group joins with fresh national survey data revealing a paradox: Americans are using AI more than ever, yet their fear of it keeps growing. He unpacks what a recent poll about revealed about AI and what types of messaging may be most effective. Plus the vivid reason he says a positive message has to be "three times as loud" to break through. And the one argument that actually moves skeptics? It has everything to do with China.

Watch Here

The Quantum Executive Order: Putting a Stake in the Ground: President Trump signed two new directives aimed at keeping the U.S. ahead in quantum computing — and accelerating the technology out of the lab and into the real world. Olivia Armstrong and Cameron Chehreh, Vice President of Field Engineering at IonQ, break down what's actually new in this strategy, why binding it with hard deadlines matters in a race against China, and how quantum fits into a much bigger push to reindustrialize America.

Watch Here

China Watch: ASML, Chinese AI Models & Polestar Ban: Channing Lee steps in for David Lin to cover a packed week out of Beijing — an explosive claim that ASML's most advanced chipmaking machine may have reached China, a new Chinese AI model outscoring a top U.S. system on a key benchmark, and Alibaba's lawsuit to get off the Pentagon's blacklist. She also unpacks how Beijing rationalizes copying U.S. technology — and why that mindset should concern American policymakers tracking the competition ahead.

Watch Here

We're live every Friday at 11 AM ET on YouTube. Subscribe so you never miss an episode — and send us your questions, comments, and feedback. Consider yourself briefed.

Washington has a bad habit of grading its most important relationships by the mood in the room. For a year now the mood between the United States and India has been foul, so the verdict seems to be that the partnership is in crisis: Washington is ambivalent to New Delhi, New Delhi is drifting away from partnership, Pakistan is the new favorite. Nearly all of this is wrong.

It is wrong because it commits the oldest sin in American statecraft, what former National Security Advisor H.R. McMaster calls strategic narcissism — the conceit that the world turns on our feelings, our deals, our news cycle. Grade a partnership by the press conference and you will misread it every time. However, we should concede the tensions first, because they’re there. But if we grade the partnership by what our two countries build when neither side is in a good mood, then a very different picture comes into focus. By that measure, and it is the only one that matters, the U.S.–India relationship is not weaker than it was twelve months ago. It is more mature and stronger. And on the pillar that counts most - our military partnership - it is stronger than at any point in its history.

The worst year in a generation

A year ago the tide ran high. Modi at the White House in February 2025, the COMPACT framework, a pledge to drive trade to half a trillion dollars by the end of the decade. It did not survive the summer. By July, Washington had hit Indian goods with a 25 percent tariff, then doubled it to 50 as punishment for India’s purchase of Russian oil. The visa doors swung shut on Indian engineers. And again and again the President claimed personal credit for ending last year’s India–Pakistan war — a claim New Delhi flatly rejects, and one that tramples the single oldest rule of Indian statecraft: no outsider mediates Kashmir.

Then the Gulf. On June 10, three Indian sailors died aboard the tanker Settebello, killed in the American blockade of Iran-linked shipping through the Strait of Hormuz. New Delhi summoned our chargé d’affaires, the External Affairs Minister called the Secretary of State. The grief and anger were national, and far more than most Americans grasp.

See it the way they see it

This is where McMaster’s other discipline, strategic empathy, earns its pay: the hard work of seeing a partner through its own history instead of our projections onto it.

You could feel it in the room in Delhi. On trade, senior Indian officials were unbothered — the President’s tariff policy is what it is, and New Delhi will trade blows on tariffs all day and think nothing of it. (Indian and American officials also noted that the U.S. Trade Representative would be visiting India days after the Forum.) On the safety of its citizens and the question of terror, the temperature changed completely. Dead sailors are not business as usual. They move the street, and in a democracy the street moves the government.

Underneath the anger sits a tempered logic: multipolarity and strategic autonomy. India has lived through bad American relations before and remembers a far bumpier era. It does not flinch and it does not bandwagon. It hedges, by design, because it is not a treaty ally and has no intention of becoming one. The framing we kept hearing was a “G-minus-two” — a reflex to find space wherever Indian interests diverge from a U.S.–China condominium. Mistake that hedging for drift and you have misread the entire country.

The agent and the partner

Which brings us to the fashionable thesis, that Washington has switched horses. The case is not frivolous. Pakistan convened the U.S.–Iran talks that no one else could or wanted to, hosting the shuttle diplomacy that produced the June ceasefire. The President called Pakistan’s army chief his “favorite field marshal,” dangled joint work on Pakistani oil and minerals, and gave Islamabad a 19 percent tariff while India was hit with 50.

But this confuses an agent with a partner. Pakistan is useful because it borders Iran and can carry a message. It was a switchboard when we couldn’t call Iran directly. Pakistan moves none of the things that structure American global strategy: not the balance of power in South Asia, not India’s trajectory, and not the one contest that organizes everything else, the strategic competition with China.

And none of this is new, which is the whole antidote to the panic. When the United States handed India the civil-nuclear deal in 2005, the founding act of the modern partnership, it announced a defense package for Pakistan in the same breath. The U.S.-India relationship has always carried that contradiction and grown anyway. Turbulence is not the exception in U.S.–India relations. It is the medium they grow through.

What got built

Now look at what was actually built during the worst year on record.

On October 31, 2025, with the tariff war at full boil, our Secretary of War and India’s Defense Minister signed the ten-year Framework for the U.S.–India Major Defense Partnership, the most ambitious accord the two nations have ever produced. Jet-engine co-production on Indian soil. Javelins and Strykers built together. American drones in Indian hands. A standing alliance for autonomous systems. Reciprocal defense procurement. Space and cyber. Hegseth called it a cornerstone of regional stability and deterrence, and for once the language was earned by the substance. The sharper point, which Indian officials make themselves, is structural: the framework is wired to run on autopilot, built deep enough to outlast whoever sits in the Oval Office or South Block.

In Delhi, Ylli recorded an interview with C. Raja Mohan, among the most consequential strategic thinkers India has produced and one of the earliest to chart the case for a U.S.–India entente. He put the argument better than we could. There is, he told us, a huge gap between the social-media discourse about this relationship and where the relationship actually sits. No country was more enthusiastic about Trump’s return in 2024 than India, though the style of the second term has produced real surprise and real heartburn. But strip out the noise, look only at the policy, and what you see is the same trajectory the two countries have traveled for twenty-five years. The lesson he drew is the one Washington keeps failing to learn: ignore the rhetoric, read the substance.

Two of his examples cut hard against the headlines. On trade, his wager is counterintuitive: that Trump’s tariff pressure may finally force through the agreement that eluded both capitals for years. In the first term, Washington stripped India’s trade preferences and the two sides couldn’t close even a mini-deal. This time the squeeze may produce the opposite — an interim deal now on the table that could carry trade toward half a trillion dollars by the end of the decade. On the H-1B fight, his counsel to New Delhi was blunt: stop taking it personally. It is an American quarrel about immigration between American firms and American politics, not a blow aimed at India, and the notion that every Indian engineer holds a standing right to move to the United States was, in his word, always an illusion. Even on Iran, where Pakistan played broker, he reads India’s interest without sentiment: a country this oil-dependent wins when the Gulf calms and prices fall, whoever takes the credit. Substance over optics, every time.

The center of gravity, meanwhile, is increasingly maritime. India fields the only resident blue-water navy in an ocean that carries two-thirds of the world’s oil, and the Hormuz tragedy was a brutal reminder of how exposed those lanes are. A relationship that lives on the chemistry of two leaders is a friendship. A relationship that co-produces engines, shares geospatial intelligence, sails integrated fleets from Malabar to Tiger Triumph, and treats that navy as the anchor of Indo-Pacific security is something else entirely. It is strategic architecture. And while the optics of renaming Indo-Pacific Command back to Pacific Command may cause hand wringing, Indian officials seemed satisfied the move did not change the Command’s theaters of operation and was meant to restore the heritage of the Command’s founding in 1947.

This is the fact the declinists cannot metabolize. Defense cooperation did not merely survive, instead it accelerated through every conceivable tension. A generation ago the glue of the relationship was sentiment. Today it is steel, software, and shared deterrence. That inversion would have been unthinkable fifteen years ago, and it is the single most important truth about where this partnership now stands.

Trade climbed the wall

Here is the number that should end the “crisis” talk. In 2025, even as the tariffs flew, two-way trade in goods and services between the United States and India hit a record — about 240 billion dollars, enough to make India a top-ten U.S. trading partner. Goods trade alone reached roughly 150 billion, with American imports from India up nearly 19 percent on the year. The United States is now India’s single largest export market and its third-largest source of foreign investment, with cumulative inflows of some 77 billion dollars. The tariff fight was loud. The trend line underneath it never stopped climbing.

And the deal that supposedly died over the summer may yet be born. As Raja Mohan argued to us, Trump’s pressure could deliver what years of polite negotiation never did: an interim trade agreement, now close, that both governments tie to the “Mission 500“ goal of half a trillion dollars in annual commerce by 2030. A relationship in crisis does not set records and chase a doubling of trade in a few short years. This one did both.

The iPhone goes to India

If you want the techno-industrial realignment in a single object, pick up an iPhone. In 2025 Apple assembled about 55 million of them in India — up 53 percent in a year — roughly a quarter of its entire global output, and it has shipped more than 50 billion dollars’ worth out of the country since joining India’s production-incentive scheme. In Spring 2025, for the first time, India out-shipped China in iPhones bound for the American market. Apple now means to build the majority of U.S.-bound iPhones in India by the end of 2026. The signature product of the U.S.–China era is becoming, assembly line by assembly line, a U.S.–India product.

The iPhone is just the visible tip of a much larger migration. India now hosts roughly 1,700 “global capability centers“ — the in-house engineering, finance, and R&D arms of multinational firms — employing close to two million people and generating some 65 billion dollars a year. About two-thirds of that demand comes from American companies. Google, Microsoft, Amazon, JPMorgan, Goldman Sachs, and Walmart run some of their largest operations on earth out of Bengaluru and Hyderabad, and more than a hundred new centers opened in the last two years alone.

This is also where the H-1B fight quietly resolves itself, exactly as Raja Mohan predicted. When Washington raised the cost of bringing Indian engineers to America, the work did not disappear — it moved to where the engineers already are. The visa wall pushed American firms to build in India rather than import from it. That is not decoupling from a partner. It is fusing with one.

The only contest that matters

The corporate migration is one half of the story; the other is what India is building for itself in the techno-industrial competition: semiconductors, artificial intelligence, critical minerals, and the resilient supply chains beneath them.

On Memos to the President, our SCSP podcast, Ylli sat down with S. Krishnan, the Secretary of India’s Ministry of Electronics and Information Technology. His framing of AI is the inverse of Washington’s anxiety. India can afford to treat AI as a force for good rather than a threat, he argues, because it has comparatively few white-collar jobs to lose and an acute shortage of the people a nation needs — doctors, nurses, judges, police. For a government that is chronically under-resourced and fiscally constrained, AI is less a job-killer than a force multiplier, a way to make the people it already has deliver far more. The goal he keeps returning to is Modi’s: a developed India by 2047.

That points to where Krishnan thinks the real edge lies — not the frontier, but the application layer. The benefits of AI accrue, he argues, only when the technology is pushed into the sectors that matter, agriculture and health and manufacturing, and even in a world of agentic workflows you still need people who understand the domain to judge the output and deploy the systems, in India and abroad. He likened it to the Y2K moment that built India’s software industry. The state is laying the foundations to match: India’s AI Mission is putting some 38,000 GPUs of compute in the hands of startups at roughly a third of the going global price, on the theory that inclusion, not hoarding, is how a billion-plus country wins. The hardware ambition is just as deliberate — India already trains close to a fifth of the world’s chip-design workforce, much of it already inside American firms, and the first phase of its Semiconductor Mission has committed some ten billion dollars to a dozen new fabs and packaging units, with a larger second round now in train. On models he is pragmatic to a fault: support both open source and proprietary models, support a score of homegrown models, and accept that for most day-to-day tasks something at sixty or seventy percent of the frontier delivers enough value. Sovereign capacity, he notes, matters for a country that has been denied technology before.

Two moments in that conversation matter most here. The first was what Krishnan said about the relationship itself: that the U.S.–India tech partnership survives changes of administration because the people-to-people ties and shared democratic values beneath it are simply too strong to ignore. That is the thesis of this newsletter, spoken by the man building India’s half of it. The second was the convergence on hard problems. A year ago New Delhi saw no need to regulate AI; today, watching how fast frontier capability is moving, Krishnan talked about putting “seatbelts” on the most powerful models before release — controlling who gets access to systems dangerous enough to find and exploit the vulnerabilities in the world’s code. When Washington and Delhi start worrying about the same frontier risks at the same time, that is not drift. That is alignment, on the only contest that counts.

Friction is the medium

A democracy that brings scale, talent, manufacturing, a blue-water navy, and strategic geography astride the Indian Ocean is not a luxury in the contest with China. It is a requirement for the resilience of the free world. We do not win this contest without India, and we should stop pretending a tariff schedule rewrites that arithmetic. Such friction between two large, proud democracies is not a sign of failure. It is the ordinary cost of serious business between sovereigns, and the medium through which they grow.

So stop reading the thermometer in the room. Read what gets built when the room goes cold. Look at what they co-produce, not what’s trending on social media.

Artificial intelligence is advancing faster than any technology before it, yet the United States still has no federal framework to govern it, leaving a patchwork of state and foreign rules that startups can barely navigate. With Europe regulating aggressively and China surging forward, the central tension is sharper than ever: regulate too much and risk falling behind, regulate too little and risk chaos. In this episode, we ask: As AI reshapes jobs, children’s lives, and the integrity of our information, can America strike the right balance between innovation and safety before it’s too late?

Teresa Carlson joins NatSec Tech to share her perspective on AI policy and the global technology race. Teresa is CEO of General Catalyst Institute, which bridges venture capital and government to advance national resilience, and previously founded and led Amazon Web Services’ Worldwide Public Sector business. Drawing on more than two decades at the intersection of big tech and government, she explains why the recent federal executive order on frontier models is only a first step, how the “Mythos moment” became a wake-up call for Washington, and why she believes the gravest risk isn’t regulation, but a failure to act with urgency.

Listen on Spotify

Listen on Apple Podcasts

As the United States and Taiwan deepen their semiconductor partnership, can they build a talent pipeline resilient enough to sustain it?

Recorded live in Orlando at the 2026 U.S.-Taiwan Semiconductor Higher Education Networking Event, on the sidelines of the NAFSA: Association of International Educators 2026 Annual Conference, host Channing Lee sits down with three university leaders shaping that future: Dr. Lin Chi-Hung, President of National Yang Ming Chiao Tung University in Taiwan; Dr. Cynthia Teniente-Matson, President of San Jose State University in the heart of Silicon Valley; and Dr. Reza Abdolvand, Professor and Chair of the Department of Electrical and Computer Engineering at the University of Central Florida. Fresh off the signing of a new memorandum of understanding, they dig past the buzzwords into what an education partnership really requires, from micro-credentials to cross-border student exchanges, why education may be the ultimate win-win in U.S.-Taiwan relations, and how each side defines success.

Listen on Apple Podcasts

Listen on Spotify

Artificial intelligence isn’t just changing how we work—it is fundamentally reshaping the rules of scientific discovery. To map out this new landscape, SCSP will host the AI+ Discovery Summit next month.

• 📅 When: July 21st

• 📍 Where: Washington, D.C.

This gathering will bring together top minds from across industry, science, and government with a singular, vital goal: ensuring the United States maintains its competitive edge at the vanguard of AI and research. From looking toward next-generation breakthroughs to diving into critical initiatives like the Department of Energy’s Genesis Mission, we will outline the strategic blueprint needed to secure America’s tech leadership.

Register for the AI+ Discovery Summit

🎙️New Episode of Memos to the President

On the margins of the India-US Forum in New Delhi, Ylli sat down for a new episode of Memos to the President with S. Krishnan, Secretary of India's Ministry of Electronics and Information Technology (MeitY), alongside SCSP Senior Advisor Dr. Sameer Lalwani.

Secretary Krishnan goes in depth on India's AI vision, the $10 billion semiconductor strategy, and how India is positioning itself as a trusted tech partner for the Global South.

Available wherever you get your podcasts.

The views expressed in this article are solely those of the author and do not necessarily represent the views of the SCSP, the ISF, or any affiliated entities.

The Robotics Data Gap: How Training Data Pipelines Will Shape the Global AI Race

The Invisible Bottleneck

Today, at a facility on the outskirts of Beijing, a young woman wearing a VR headset and an upper-body exoskeleton is folding a pair of shorts. She will repeat this motion hundreds more times today. Every fold is recorded by joint torque sensors, multimodal cameras, and LiDAR arrays — every wrist rotation, every finger’s pressure, every adjustment of the fabric captured as structured data. Tomorrow, a humanoid robot will attempt to replicate her movements. She is one of hundreds of what Chinese media have dubbed “cyber laborers” — a new class of infrastructure that the rest of the world is only beginning to understand.

The global race to build autonomous robots has captured the attention of policymakers and investors across major economies. Governments debate chip export controls, think tanks analyze hardware supply chains, and venture capitalists pour billions into humanoid startups; yet the bottleneck that will most determine which nations lead in this technology receives almost no policy attention: the physical AI1 training data pipeline.

The disconnect is fundamental. Language models are trained on the internet — trillions of words at near-zero marginal cost. However, a robot cannot learn to fold clothes from Wikipedia. It requires teleoperated demonstrations, sensor recordings, and manipulation sequences — all physically collected through human labor on hardware rigs costing $50,000 to $150,000 each, and yielding fewer than 200 demonstrations per worker per day.

The global AI data annotation market, valued at $4.89 billion in 2025, is projected to reach $17.1 billion by 2030 — with robotics emerging as its fastest-growing segment. Micro1 has contracted thousands of workers across more than 50 countries, paying $15 per hour to strap iPhones to their heads and film household chores; DoorDash has launched a service called “Tasks” leveraging its couriers to record robot training data; Scale AI has logged more than 100,000 hours of in-house collection in its San Francisco labs.

The deeper insight missing from current policy discussions is that teleoperation facilities are not merely cost centers — they are infrastructure. Just as TSMC’s plants became indispensable bottlenecks in the semiconductor supply chain, dedicated robotic data collection facilities will become critical bottlenecks in the physical AI supply chain. Whoever establishes the trusted, well-managed pipelines will hold leverage comparable to a semiconductor fab. The question is not whether this infrastructure will be built — it is where, by whom, and under what standards.

Two natural questions arise. First, why not synthetic data, given projections it could comprise the majority of AI training data by 2030? Synthetic generation does scale demonstrations aggressively — Nvidia’s data factories augment a single captured motion into millions of variants — but it amplifies a real corpus; it cannot originate one. You cannot simulate contact dynamics, friction coefficients, or sensor noise patterns you have never observed. Every synthetic variant traces back to a real seed, which is why Nvidia identified demonstration data as the critical bottleneck for physical AI at GTC 2026 even while projecting heavy synthetic augmentation downstream. Second, why not use the iPhones strapped to gig workers’ heads? Phones record what a fold looks like, but cannot capture the joint torque, force feedback, depth, or exoskeleton kinematics that contact-rich manipulation depends on. A model trained on smartphone video learns the appearance of a successful fold; it cannot learn the proprioceptive signal that distinguishes a secure grip from a slip. For visually loose tasks, lower-cost capture is plausible; for the dexterous manipulation that defines real-world robotic value — household chores, eldercare, assembly — the sensor stack is non-negotiable.

This is the “robot data gap”: whoever can build the infrastructure to collect, standardize, and manage this data at scale will hold the key to the next era of industrial automation. Currently, one country is clearly building it fastest.

China’s National Robotics Data Strategy

China recognized the robot data gap as a strategic priority earlier and more comprehensively than any other country, offering both a model worth studying and a demonstration of the national ambition required for data supremacy in a critical emerging field.

While at Tsinghua University, I witnessed this transformation firsthand. In a fall 2025 seminar, a professor advising the Ministry of Industry and Information Technology’s (MIIT) committee on humanoid robot standardization outlined the government’s data collection roadmap: “embodied intelligence” — a term that appeared almost nowhere in Chinese policy documents prior to 2023 — has become a top research priority and a leading hiring category. Recruitment posters for “cyber laborer” positions now appear alongside traditional engineering roles on Tsinghua job boards.

In December 2025, MIIT appointed a 65-member standardization committee for humanoid robots and embodied intelligence — including the founders of Unitree and Zhiyuan Robotics (智元机器人), Huawei executives, and Tsinghua scholars. By March 2026, MIIT released its first national standard system, covering the entire lifecycle from data acquisition formats to safety protocols.

The 15th Five-Year Plan (2026–2030) elevated the field further, designating robotics one of eight “strategic emerging industries” and dedicating a section to embodied intelligence among the top ten “new industry tracks” — triggering mandatory coordination across central ministries and state financial institutions. Beyond $20 billion in direct subsidies, an NDRC guidance fund plans to inject $137 billion into AI and robotics over the next two decades.

The infrastructure surrounding Beijing offers a concrete lesson for any nation serious about physical AI. In 2025, more than 40 government-backed robot training centers were established across China, with approximately 20 already fully operational. The Beijing Humanoid Robot Data Training Center spans over 10,000 square meters, employs hundreds of workers, and covers 16 task categories from automotive assembly to elder care. In Zigong, Sichuan, a 6,000-square-meter facility opened in January 2026 is capable of generating three million high-quality data entries annually — comparable to the entire scale of the Open X-Embodiment dataset, which aggregated the work of 22 institutional contributors.2

Beyond standardization committees and industrial parks, Chinese firms are converging on a vertically integrated robotics stack, from sensors and rigs to data and foundation models.3 At the May 2026 JP Morgan China Summit, the depth of talent and capital behind Chinese physical AI was on open display.4

What makes this distinct is not the scale of labor but the degree of state coordination behind it. The PRC has done what no other market would: pre-financed and standardized the physical infrastructure of an entire training pipeline before downstream demand justified the investment.

America's Missing Strategy

Despite world-class robotics research institutions and a vibrant startup ecosystem, the United States lacks a national robotics data strategy. As the Special Competitive Studies Project’s (SCSP) National Robotics Strategy report5 lays out, China has spent the better part of a decade executing a coordinated national robotics strategy — accelerating adoption, scaling manufacturing, and securing its supply chain — while America has yet to mount a comparable effort.

The investment gap reflects this. The ARM Institute — America’s flagship public-private robotics accelerator — operates on a $30 million budget. In 2025, even Tesla — among the most prolific U.S. humanoid producers — is estimated to have shipped only 150 humanoid robots, while Chinese firms took the global lead. The gap reflects coordination and public investment, not lack of American talent or technology.

Washington is beginning to take notice. In June 2026, Senators John Hickenlooper, Dave McCormick, Martin Heinrich, and Todd Young introduced bipartisan legislation to establish a National Commission on Robotics tasked with developing recommendations to strengthen U.S. competitiveness, supply chains, workforce development, and national strategy in robotics. The American Security Robotics Act, introduced in March 2026, would restrict federal procurement of foreign-manufactured humanoid robots; the White House is reportedly drafting an executive order on robotics; Commerce Secretary Howard Lutnick has convened meetings with the CEOs of Apptronik, Boston Dynamics, and Tesla. These initiatives are largely reactive — focused on strategy development and procurement restrictions rather than on the proactive data infrastructure U.S. robotics companies need. The critical question is not what to block, but what to build.

Meanwhile, China is actively shaping international standards, leading the International Electrotechnical Commission’s (IEC) standard-setting efforts, such as for elder-care robots, and establishing domestic norms that could become the global default. This is what any leading producer would do. But it means other nations — including the United States — must proactively engage, or accept standards they played no part in designing.

A Three-Pillar Policy Framework

The robotics data gap requires a coordinated response. I propose a three-pillar framework addressing strategy, labor, and diplomacy.

Pillar One: A National Robotics Data Strategy

The United States should establish a coordinated national strategy for robotics training data — not by mimicking China’s centralized approach, but by drawing on the federal public-private partnerships that have historically driven American dynamism.

At the federal level: establish a National Robotics Data Initiative within the Department of Commerce to coordinate data collection standards across the Department of War, Department of Health and Human Services, and The National Institute of Standards and Technology; fund shared robotics data infrastructure (teleoperation facilities, sensor suites, annotation tools) through competitive grants; establish common data formats and interoperability standards to prevent fragmentation from proprietary pipelines; and launch a robotics data accelerator (modeled on the DARPA Challenges) matching funds for companies building scalable teleoperation operations.

At the industry level: form a robotics data consortium (modeled on the Semiconductor Research Corporation) where rival firms jointly invest in pre-competitive data infrastructure while retaining proprietary advantages in model training; invest in dedicated teleoperation facilities staffed by a specialized workforce; and adopt and expand open standards such as Nvidia’s Isaac Teleop and the Open X-Embodiment framework.

Pillar Two: International Labor Standards for Teleoperators

The emerging teleoperation workforce requires protections existing labor laws fail to provide — not as an ethical add-on, but because labor standards and data quality reinforce one another: better-trained workers in structured environments produce better data. This can also distinguish an American approach to teleoperated robotics data from the PRC’s approach.

At the federal level: establish a new occupational classification of “Data Demonstration Workers,” recognizing their role in generating training data for physical intelligence; institute minimum standards for wage transparency, including disclosure of how collected data will be used and its expected commercial value; and mandate that platforms recruiting international teleoperators meet labor standards of both the worker’s country of residence and the country where the commercial application is deployed.

At the international level: the International Labour Organization (ILO) should convene a working group on AI data labor and expand its framework for digital labor platforms to cover demonstration work; multilateral development banks should condition loans to the robotics industry on adherence to minimum standards for data workers.

Pillar Three: International Data-Sharing Agreements

No single country — including China — can independently gather the full diversity of data required to train robots; the diverse environments those robots will operate in demand demonstration data drawn from all of them. Workable precedents exist: ISO and IEC robotics safety standards (ISO 10218 for industrial robots, ISO 13482 for personal care robots) have for years brought American, European, Japanese, and Chinese engineers to the same technical tables, and the Bletchley (2023) and Seoul (2024) AI Safety Summits demonstrated that even strategic competitors can share evaluation methodologies on narrowly bounded technical questions.

At the government level: negotiate mutual recognition agreements for robot data standards, ensuring demonstration data collected in one country meets quality requirements for use in another; and pursue narrowly scoped data-sharing arrangements with China on robot safety, eldercare, and disaster-response applications. The IEC’s existing elder-care robot working group — which China leads but in which U.S. and European representatives also participate — is a working example of how technical cooperation can coexist with strategic competition.

At the standards-body level: actively participate in IEC and ISO data governance processes, collaborating rather than competing with China and other leading nations to develop interoperable global frameworks; and develop data provenance standards to establish a trusted supply chain for robot training data — analogous to the “Trusted Foundry” concept in semiconductors.

The Stakes

The robot data gap is not a distant policy issue — it is a present reality. China’s 15th Five-Year Plan projects that the market for embodied intelligence will reach 400 billion RMB ($56.5 billion) by 2030 and 1 trillion RMB by 2035. The training data infrastructure being built today — in China, and eventually elsewhere — will accumulate over time, creating a first-mover advantage increasingly difficult to replicate.

The first phase of the AI revolution was built on internet-scale text data and world-class research institutions. The second phase — physical AI — will be determined by who can master the collection, standardization, and governance of embodied data to deploy physical robotics platforms at scale.

The robotics data divide can be bridged. But only if policymakers recognize that the most impactful supply chain of the AI era is not silicon but the human demonstrations that teach machines how to move.

This spring, SCSP delved into agentic AI, including discussions with leading experts from industry, academia, and government to understand where agentic AI stands — and where it’s headed. The message is consistent: the technology is moving fast, and our institutions need to be asking the right questions.

Below is what we’ve learned about agentic AI— and what it demands of us.

The strategic competition dimension

Any honest assessment of agentic AI must begin here because we must simultaneously harness the benefits while protecting against threats this technology presents.

Beijing understands what is at stake. China’s approach to governing agentic AI has been deliberate and instructive: existing regulation, standards and specifications, rather than agentic AI specific hard regulation. This enables China to preserve maximum flexibility for deployment while avoiding premature constraint on domestic development. Whether by design or adaptation, its approach has allowed it to preserve flexibility for deployment while avoiding constraints that could slow domestic development.

The competitive threat is not simply that China may field agentic AI systems before we do. It is that adversaries will deploy agentic systems precisely in the domains where our governance is weakest. Agentic systems that can act across networked environments are not merely economic tools. They are potential instruments of coercion, espionage, and influence. An agent that can navigate complex bureaucratic systems, identify exploitable vulnerabilities, and act without leaving a clear attribution trail represents a qualitative expansion of adversarial capability.

And these capabilities can improve fastest exactly where it matters most to us. In cyber operations, influence campaigns, and economic intelligence collection, adversarial feedback loops are tight and the incentive to improve is relentless. The United States should expect that in these domains, adversarial agentic capabilities are advancing as fast as or faster than our own.

The competitive reality is the reason for this assessment. The countries that succeed will not be those that merely develop agentic AI systems first, but those that learn how to deploy, govern, and scale them effectively.

We must focus on governing the agentic capabilities

The term “agent” has been overloaded into a catch-all. Instead of debating the term, policymakers should focus on agentic capability.

The operationally relevant distinction is this: a tool executes what you instruct, step by step. An agent is handed a goal and determines the execution steps itself. The difference is not cosmetic. It is the difference between a calculator and a subordinate, between automation and delegation.

What makes agentic systems genuinely transformative is not the AI model that is the engine. It is the scaffolding built around it:

• Connectors that bridge the model to real-world infrastructure (e.g., email, booking systems, financial platforms);

• Memory that allows the system to learn and adapt across interactions over time;

• Planning capabilities that decompose objectives into smaller tasks, identify failure, and route around obstacles without human intervention;

• Permission structures that define what the system can access and act upon; and

• Guardrails that determine what the system will refuse to do (e.g., spending limits or human sign-offs).

This architectural reality has a direct policy implication: governance aimed at “the model” misses where the “agentic” capability and real risk live. Two different agentic systems running identical models can behave entirely differently depending on their scaffolding. Regulating the engine while ignoring the keys, the fuel, and the road is not a serious strategy.

Where the technology is heading

Progress in agentic AI will not be uniform. It will follow a heat map, not a tide. AI improves fastest where verification of right and wrong answers is possible. Where outcomes are measurable and feedback is rapid (e.g., software engineering, mathematical reasoning, structured data analysis) capabilities are already mature and accelerating. Where success is harder to define, the technology remains fragile.

Near term (zero to twelve months)

Agentic systems will move aggressively into paperwork-intensive professional domains: legal drafting, tax preparation, financial analysis, and regulatory compliance. This is not simply faster processing of existing work. These are domains where agentic systems will begin fundamentally restructuring how professional judgment is exercised and verified.

The next leap in agentic capabilities is orchestration, systems built not to perform tasks themselves, but to direct, supervise, and coordinate networks of specialized agents in parallel towards a shared goal. The capability gain is not linear, but compounds as orchestrators begin supervising other orchestrators. This scale expands both the opportunity and risk surface dramatically. And the speed and complexity of actions will exceed what human oversight can meaningfully track under current governance frameworks.

Medium term (two to five years)

Agentic systems will enter physical environments through robotics and embodied AI, starting in controlled industrial settings. Healthcare represents a distinct and highly consequential use case: agentic systems stand to transform, for example, records infrastructure, clinical decision support, and laboratory research, with implications for cost, access, and liability that existing regulatory frameworks are not designed to handle.

The one trend to watch above all others

AI is beginning to help build better AI. The self-accelerating loop, in which AI capability improves AI development, is no longer theoretical. If this feedback cycle compounds, the pace of capability development will outrun even optimistic projections. This is the single development that policymakers must track above all others.

The accountability deficit

The central finding is that governance is losing ground to technology. Three failures are already visible.

• No one can say who is responsible. When an agent acts on your behalf, and spawns sub-agents to complete parts of the task, there is currently no reliable mechanism for establishing who authorized what, or who bears responsibility when something goes wrong. The Air Canada tribunal ruling, which held the airline liable for misinformation provided by its chatbot, signals that institutions will likely not be able to disclaim responsibility for their agents’ actions. But the legal and technical architecture to operationalize that principle does not yet exist.

• Evaluations ask the wrong question. Current assessment frameworks ask whether an agent completed a task. They do not necessarily ask whether it did so safely, whether it could be manipulated into doing something else, or whether it caused harm that was not immediately visible. Task completion, thus, is not a sufficient measure of trustworthiness.

• Privacy risks are structural, not incidental. Because agents build persistent models of the individuals they serve (e.g., accumulating preferences, behaviors, patterns, and inferences), they generate sensitive profiles. Existing privacy frameworks were not designed for this model of continuous, inferential data accumulation.

The backlash risk

A concern that deserves direct mention. A cascade of high-profile failures, or a single catastrophic one, could produce a political and public reaction that sets back legitimate, highly beneficial applications of the technology for years. The precedent is self-driving vehicles, which stalled for roughly six years after a 2018 fatality in Tempe, Arizona.

The lesson is not that caution is always correct. It is that the absence of serious governance creates the conditions for the kind of failure that produces overcorrection. Responsible speed and responsible governance are not in tension, they are mutually reinforcing. Countries that deploy recklessly invite backlash. Countries that govern too timidly cede the field to those who will not.

Workforce and institutional capacity

A strategic assessment that ignores the human capital dimension is incomplete. The number of people who understand agentic AI deeply enough to govern it well is small, unevenly distributed, and concentrated outside government.

Congress, regulatory agencies, and the national security apparatus cannot govern what they do not understand. Closing this gap is not a secondary priority. It is a prerequisite for everything else on this list.

What policymakers should do now

The following are not aspirational goals. They are minimum necessary actions for a country that intends to remain competitive and to govern this technology responsibly.

• Resource and codify CAISI’s agentic evaluation mandate. The capability already exists. The Center for AI Standards and Innovation (CAISI) conducts adversarial evaluation of authentic systems, under its AI Action Plan taskings. The gap is durability and scale. CAISI operates on a fraction of the budget its mandate implies, and its statutory footing is fragile. CAISI needs to be fully funded to the level of its agent-AI taskings and so the capability survives beyond any single administration.

• Use federal procurement as a forcing function. OMB and GSA should require that any agentic system sold to the federal government include a tamper-evident action log, or some similar mechanism, that records who authorized what, which sub-agents were generated, and what each accessed and the action taken. This sets a de facto national standard.

• Establish accountability architecture. Departments and agencies should adopt a method whereby every deployed agent maps to an identifiable legal person who holds liability, and that liability cannot be delegated away to sub-agents. The model already exists in financial services, where a named individual is accountable for an automated system’s conduct.

• Close the talent gap outside CAISI. CAISI has dedicated hiring authorities and is scaling its technical workforce, but the rest of the government is not. The agencies that will have to govern agentic AI in practice (e.g., the sectoral regulators, the national security apparatus, and the congressional committees) lack comparable access to AI-literate personnel. Fellowships, rotational programs, and direct-hire authorities should be extended to them at the pace the technology demands.

• Engage allies before adversaries arbitrage our governance weaknesses. Governance built in isolation is governance adversaries can route around. The United States should work with partners on common standards for the domains where agentic risk is most highly consequential (e.g., healthcare, cyber, finance, critical infrastructure).

Conclusion

Everything we’ve learned about agentic AI capabilities validates the same observation: the technology is moving faster than our ability to govern it. That is a warning, not an excuse for paralysis. The United States has navigated strategic competition before (e.g., nuclear weapons, space, the early internet) and has done so best when it combined urgency with seriousness, speed with deliberation.

Agentic AI is not a technology to be feared or deferred. It is a capability to be understood, shaped, and governed. The institutions that accomplish this will help determine not only their own competitive position, but the character of the environment in which this technology operates globally.

The bottleneck is not the AI. It is our willingness to govern it with the same urgency we bring to building it.

Join SCSP at the AI+ Discovery Summit! 💡

AI isn’t just changing how we work—it is fundamentally reshaping the rules of scientific discovery. To map out this new landscape, SCSP is excited to host the AI+ Discovery Summit next month.

• 📅 When: July 21st

• 📍 Where: Washington, D.C.

This gathering will bring together top minds from across industry, science, and government with a singular, vital goal: ensuring the United States maintains its competitive edge at the vanguard of AI and research. From looking toward next-generation breakthroughs to diving into critical initiatives like the Department of Energy’s Genesis Mission, we will outline the strategic blueprint needed to secure America’s tech leadership.

Register for the AI+ Discovery Summit

If you only have two minutes today, watch these:

Iran: Lose the War, Win the Peace?: A memorandum of understanding between Washington and Tehran is circulating — and the early read from our experts isn't reassuring. Former Ambassador Puneet Talwar and former NSC Iran director Joel Rayburn dissect a deal that, in their telling, rewards a militarily devastated regime for simply reopening the Strait of Hormuz. They explain why the war's founding assumptions collapsed, where the next 60 days of nuclear talks could break down, and why one of them still calls this "the least bad alternative" the President had.

Watch Here

The Philippines vs. China: "How Do You Provoke Someone With No Right to Be There?": Philippine Secretary of National Defense Gilberto Teodoro joins from California for one of the sharpest segments yet. He breaks down China's two-pronged playbook of co-option and coercion, the AI-driven disinformation aimed at Filipino resolve, and explains why China's claim that the Philippines provokes it simply doesn't hold up.

Watch Here

The White House vs Anthropic: 90 Minutes to Go Dark: Last Friday, the White House did something it had never done before: citing national security, it used export-control authority to restrict access to Anthropic's most powerful model. Anthropic shut the model down within 90 minutes. Rama Elluru and Jenilee Keefe Singer unpack the legal mechanism behind the move, the gap it exposed between what technologists and government each understand about frontier models, and who reviews these models and who can actually enforce.

Watch Here

The G7: Better Vibes, Bigger Pressure: For the first time since last year's summit — which ended with no joint statement at all — G7 leaders agreed on common language. Joe Wang breaks down what actually moved: a notable shift on Russia as falling energy prices open room to tighten pressure, a fresh commitment to accelerate military aid to Ukraine, Macron's pitch for a critical-minerals alliance with its own secretariat, and a tech-leader working lunch on AI that revealed more about what leaders don't yet grasp than what they agreed on.

Watch Here

China Media Watch: Gavin Finkle returns with our AI-powered scan of Chinese state media — 40 high-risk entries this week, distilled to three. The throughline: Beijing is no longer treating these as separate stories. From how CGTN is reframing the Iran outcome, to a "computing power network" China is presenting as national infrastructure, to a cluster of Indo-Pacific moves tied together by one message — defensive, but uncompromising.

Watch Here

We're live every Friday at 11 AM ET on YouTube. Subscribe so you never miss an episode — and send us your questions, comments, and feedback. Consider yourself briefed.

🎧 This week on Intel at the Edge

In today’s episode of Intel at the Edge, former NSA Director General Tim Haugh and former GCHQ Director Sir Jeremy Fleming sit down for a rare, candid conversation about one of the most consequential intelligence relationships in the world.

From the wartime roots at Bletchley Park to the age of AI and quantum computing, they explore how the US-UK signals intelligence alliance has evolved over 80+ years and discuss what it must do to stay ahead of an increasingly complex threat landscape.

Available wherever you get your podcasts.

Apple Podcasts

Spotify

Launching the 2027 International Strategy Forum Application

Founded by SCSP Chair and former Google CEO Dr. Eric Schmidt and co-chaired by CNN host Fareed Zakaria, ISF is a year-long, non-residential fellowship program that empowers rising global leaders aged 25 to 35 to explore the future of geopolitics, technology, and democracy. ISF brings together individuals from diverse fields, including government, academia, deep tech, the military, journalism, and the arts. In 2024, ISF was integrated into SCSP and has become a signature aspect of SCSP’s global network of leaders and experts.

The ISF fellowship year is organized around a series of in-person convenings alongside virtual programming, enabling Fellows to engage with each other and the broader ISF and SCSP communities. Our goal is to forge a network of next-generation problem solvers with extraordinary potential in geopolitics, innovation, and civic leadership, who can work across sectoral and national lines to collectively tackle the challenges of a changing world order.

The 2027 Application

If you or someone you know is eligible and interested in learning more, please find the application here.

Competitive applicants will have the following characteristics:

• Demonstrated leadership potential.

• Recognized expertise in their professional field.

• Clear commitment to ISF’s values of purpose, collaboration, curiosity, and empathy.

• Strong devotion to building a network of interdisciplinary leaders working to solve the world’s hardest problems.

The application deadline is July 31, 2026. Selected applicants will be invited to participate in interviews in the months that follow.

If you have any questions, please don’t hesitate to contact the ISF team at isfteam@scsp.ai with the subject line “ISF in 2027.”

Apply today