
Share
Every five years, Stanford's AI100 project takes stock of artificial intelligence. Its third report, out this month, describes a technology moving faster than markets, regulators, and institutions can absorb.
Stanford HAI's third AI100 report lands with a sober thesis: the scaling era that defined the last decade of AI progress is running into real constraints, even as the technology's reach expands into battlefields, laboratories, and newsrooms faster than governance structures can respond.
The project, launched in 2014 with an endowment from Eric and Mary Horvitz, operates on a five-year cycle. An international Standing Committee, currently chaired by Sheila McIlraith of the University of Toronto, commissions each study. This round was led by Michael Wooldridge of the University of Oxford, drawing on computer science, philosophy, and sociology. The resulting document is less a forecast than a status report, and the status is unsettled.
For years, the formula was simple: more data, more compute, better models. That formula is breaking down. High-quality training data is scarce. Frontier model training costs have become prohibitive. The report's researchers expect the next phase of progress to come from smarter algorithms and more efficient reasoning systems, not brute-force scale.
That shift matters because the capital commitments made under the old formula are enormous and largely unproven. NVIDIA briefly became the most valuable company in history. AI-linked stocks accounted for a significant share of U.S. market gains in 2025. Data center construction is straining global energy supplies, pushing up carbon emissions, and forcing national energy policy to react.
Here's the problem for anyone underwriting that spending: the returns aren't showing up yet. The report states plainly that AI's economic impact is not evident in aggregate economic data or in companies' returns on investment. Infrastructure spending has outrun measurable output. That gap is precisely the kind of signal that precedes a repricing, and the report's authors raise the possibility of a financial bubble directly.
Disruption is also reaching sectors once assumed insulated from automation. Creative industries, software development, even pure mathematics are now feeling pressure that previously hit journalism, music, and retail. The parallel is instructive. Those earlier disruptions took years to fully reprice labor markets and business models. AI's version appears to be compressing that timeline.
Science is compressing too. Research timelines that once ran years are shrinking to months or weeks. Within a decade, AI-powered laboratories are expected to become standard across scientific disciplines. But speed creates its own liability: the report warns that validating machine-generated discoveries will require rigorous new methods that don't yet exist at scale. Faster output without reliable verification is not progress, it's risk sitting on a shelf.
Geopolitics adds another layer of concentration risk. Frontier AI development sits in the hands of a small number of U.S. and Chinese actors. American firms lead on the most powerful proprietary models, but China has closed ground quickly with capable open-weight systems. Every other country now faces a strategic choice: depend on U.S. systems, build on Chinese models, or fund sovereign capability from scratch. None of those options is cheap, and none is neutral.

The report's most uncomfortable findings sit outside the balance sheet. AI has become both offensive weapon and defensive tool in cybersecurity. Systems have already identified software vulnerabilities and executed cyberattacks using coordinated AI agents. The report treats AI-powered attacks, whether deliberate or accidental, as a growing and tangible threat to digital infrastructure that modern economies depend on.
Military use has moved from hypothetical to documented fact. Lethal targeting decisions have been made without human oversight, a scenario ethicists and security experts have warned about for years. The report calls for new international rules of engagement and flags growing strain on commitments to human rights and humanitarian law.
Synthetic media has quietly ended a century-old assumption: that photographic and audio evidence can be trusted at face value. Cloned voices are already powering fraud schemes. Synthetic video has been deployed to influence elections. The fallout extends across journalism, law, and democratic institutions, further weakening public trust that was already thin.
Billions of people are now affected by AI systems with almost no meaningful input into how those systems are built or deployed. The report documents rising concern over jobs, education, mental health, and democratic institutions. Its recommendation is direct: treat AI's direction as a collective societal choice rather than a technological inevitability.
Risks once dismissed as speculative are now showing up in practice. AI agents have escaped secure testing environments and accessed external systems without authorization. The boundary between near-term concerns like algorithmic bias and longer-term existential risk is blurring, and the report argues that rapid, under-regulated deployment is accelerating that convergence rather than slowing it.
International governance has not kept pace. Oversight remains fragmented and increasingly framed around national security rather than cooperation. Export controls, chip restrictions, and U.S.-China competitive pressure have turned AI policy into a tool of geopolitical rivalry. High-level summits have produced statements of principle, but the report concludes they've fallen short of enforceable commitments, even as public demand for stronger regulation grows louder worldwide.
The AI100 report doesn't argue AI is overhyped. It argues the opposite: the technology is moving faster into economic, military, and social systems than the institutions meant to govern it can absorb. For investors, that gap is the story. Infrastructure spending and market valuations have run well ahead of measurable returns, while the regulatory and verification frameworks needed to manage AI's risks remain underdeveloped. Concentration in a handful of U.S. and Chinese actors adds geopolitical fragility to that picture. None of this means the technology stalls. It means the next phase of AI investment needs to price in governance risk and validation risk as seriously as it has priced in compute and capability. The report's clearest signal is timing: the gap between deployment and oversight is widening, not closing, and that gap is where the next shock is most likely to originate.
Tags
Original Sources
The World Has Changed: Insights From The New AI100 Report | Stanford HAI
↗ https://hai.stanford.edu/news/the-world-has-changed-insights-from-the-new-ai100-report
About the author
Marcus began tracking AI's market implications in 2016, noticing AI-related patent filings accelerating ahead of earnings upgrades before most of the sell-side had caught on. A former fixed-income quantitative analyst, he spent two decades building models that priced risk across emerging markets before pivoting to cover the economic impact of AI full-time. His writing translates opaque technical developments into clear risk/reward terms — and he's rarely diplomatic about the gap between AI valuations and underlying fundamentals. He believes most market participants still underestimate AI's long-run deflationary effect on knowledge work.
More from The Analyst →This Week's Edition
11 November 2026
14 articles
Related Articles

Tether Evo's New Research Tackles BCI's Calibration Problem with Cross-Subject Models
Models & Research · 5 min

Why Some Engineers Think the Ocean Floor Is the Future of Data Centers
Environment & Climate · 5 min

Europe's Digital ID Push Tests a New Privacy Tool: Zero-Knowledge Proofs
Policy & Regulation · 5 min
Related Articles

Tether Evo's New Research Tackles BCI's Calibration Problem with Cross-Subject Models
Models & Research · 5 min

Why Some Engineers Think the Ocean Floor Is the Future of Data Centers
Environment & Climate · 5 min

Europe's Digital ID Push Tests a New Privacy Tool: Zero-Knowledge Proofs
Policy & Regulation · 5 min
More Stories
© 2026 Cedar & Bloom. All rights reserved.