THE AI POWER RANKING · CURRENT · 2026-09-20
WHO ACTUALLY RUNS AI?
The AI power ranking of the 50 people with the most power to shape what AI becomes next.
YOU'RE GOING TO HATE AT LEAST ONE OF THESE.
This is an editorial ranking, not a formula. We use the same lenses for every person, then make the call. No fake 94.7 scores.
The numbered list covers industry, research, capital, and infrastructure figures. Government power is reported through concrete offices, authorities, and actions rather than folded into this numerical ranking.
LEATHER JACKET, ALL THE CHIPS
Nvidia sits underneath much of the frontier race. Chips, CUDA, systems, and data-center economics make Huang the person whose roadmap can change everyone else's plans.
FIRED. BACK IN FIVE DAYS.
OpenAI combines frontier models, ChatGPT distribution, huge capital needs, infrastructure alliances, and an unusual ability to make the rest of the industry respond.
CALM FACE, GEMINI EVERYWHERE
Alphabet can move AI through Search, Android, Cloud, Gemini, and Google DeepMind at once. Pichai controls one of the broadest distribution and research stacks in technology.
COPILOT IN EVERY OFFICE
Azure, Windows, Microsoft 365, Copilot, and the OpenAI relationship give Nadella enormous leverage over how advanced AI reaches companies and workers.
PROTEINS FIRST. AGI NEXT?
Hassabis leads Alphabet's most important frontier-research organization while also shaping AI-for-science strategy, giving him unusual reach across models, research talent, and scientific applications.
CLAUDE AND THE END TIMES
Anthropic turned Claude into a frontier competitor while tying itself deeply to hyperscale infrastructure. Amodei also helps set the industry's argument about speed, safety, and control.
ZUCK'S AI TALENT SHOPPING SPREE
Meta can pair enormous compute spending with billions of social-product users and an open-model strategy designed to reshape the economics of the model layer.
GROK, FEUDS, SUPERCOMPUTERS
xAI combines a frontier-model effort with X distribution, aggressive infrastructure spending, and Musk's ability to move capital, talent, and attention across companies.
AWS GETS PAID EITHER WAY
AWS is one of the physical and commercial foundations of the AI boom. Amazon's Anthropic alliance, custom chips, Bedrock, and Nova give Jassy influence across the infrastructure layer.
HEDGE FUND → DEEPSEEK
DeepSeek changed assumptions about who can build frontier reasoning systems and at what cost. Liang now sits at the center of one of the most consequential non-U.S. model labs.
ONE GIGAWATT FOR A STARTUP
Thinking Machines moved from new-lab curiosity to infrastructure-scale ambition, including a multi-year Nvidia partnership targeting at least one gigawatt of next-generation systems.
FROM IPAD TO THE THRONE
Ternus became Apple CEO in September 2026. That hands him one of the world's largest device ecosystems, a custom-silicon stack, and enormous power over how consumer AI reaches hardware.
PARIS VS. SILICON VALLEY
Mistral has become Europe's most visible frontier-model challenger, with fresh capital and a strategy spanning models, enterprise software, and sovereign AI infrastructure.
DEEPMIND → INFLECTION → MICROSOFT
Microsoft AI puts Suleyman over consumer AI products and frontier-model work inside one of the world's largest software and cloud companies.
SUPERINTELLIGENCE, NO DISTRACTIONS
Safe Superintelligence gives Sutskever a focused frontier lab, while his technical reputation still carries unusual weight in recruiting and in arguments about what advanced systems become.
CLOUD MONEY, ROBBER-BARON ENERGY
Bezos no longer runs Amazon day to day, but he remains executive chair of the company that owns AWS and retains the personal capital to make unusually large long-horizon technology bets.
MICROSOFT RESEARCH → QWEN
Alibaba lists Zhou as chief AI architect and head of its Tongyi foundation-model division, putting him where one of China's largest cloud platforms meets the Qwen model family.
GOOGLE'S BACKBONE, NOW GEMINI
Dean is Google's chief scientist across Research and DeepMind, after decades building the distributed systems and machine-learning infrastructure beneath Google's AI stack.
IMAGENET, NOW IN 3D
Li combines scientific authority, Stanford institutional reach, and World Labs' push into spatial intelligence, giving her influence across both research direction and company building.
BUILT IT. WON NOBEL. GOT WORRIED.
Hinton does not control a giant company, but his foundational scientific status and public warnings give him unusual agenda-setting power over how advanced AI is understood and debated.
CEO OUT. CHAIR STILL IN.
Cook left the Apple CEO job in September 2026 but became executive chair. He remains attached to one of the industry's most valuable distribution, silicon, and device platforms.
A PULL-THE-PLUG LINE FOR AI
Schmidt connects a Google-era fortune, science philanthropy, national-security institutions, and AI policy debates, giving him influence far beyond any single operating company.
LLMS AREN'T THE ENDGAME
LeCun left Meta to build Advanced Machine Intelligence around world models, then raised more than $1 billion to pursue an alternative to the industry's dominant language-model path.
GODFATHER BUILDS A SAFETY LAB
Bengio pairs foundational research stature with international safety work and LawZero, giving him a durable voice in the technical and institutional response to frontier AI risk.
HUGGING FACE, OPEN EVERYTHING
Wolf helped turn Hugging Face into a central distribution and collaboration layer for open models, datasets, and tooling used across the AI research and developer ecosystem.
META OUT. MISTRAL IN.
Lample co-founded Mistral after helping build Meta's language-model research, and now helps steer the science behind Europe's largest independent frontier-model company.
MISTRAL'S ENGINE ROOM
Lacroix is the engineering co-founder and CTO behind Mistral's rapid expansion from open-weight models into enterprise systems and sovereign compute.
BILLION DOWNLOADS, THEN HE LEFT
Lin helped lead Qwen through its explosive open-model expansion before leaving Alibaba in 2026, giving him lasting influence through one of the world's most widely distributed model families.
TEACHING QWEN MANNERS
Yu leads work on Qwen post-training and alignment, the machinery that turns raw pretrained systems into useful instruction-following and reasoning models.
QWEN LEARNED TO CODE
Hui has become a prominent Qwen researcher across coding models, reasoning, and open software-engineering agents, giving him leverage where models increasingly write the software around them.
DEEPSEEK STARTED WITH CODE
Guo helped establish DeepSeek's reputation in code intelligence and reasoning research before leaving the lab in 2026, carrying technical influence from one of the industry's biggest disruptions.
VIBE CODING GOT ITS NAME HERE
Karpathy has moved between OpenAI, Tesla, research, education, and independent building. His unusual power is making frontier ideas legible and usable to working engineers.
SOUTH PARK + SEMICONDUCTORS
SemiAnalysis turns chips, data centers, supply chains, and facility-level research into information that executives and investors use to understand the physical AI race.
"I THINK WE ARE ALL GOING TO DIE"
Yudkowsky helped build the modern intellectual ecosystem around alignment and existential risk, giving him agenda-setting power far beyond the size of his institutions.
THE SOCIAL DILEMMA, NOW WITH AI
Harris turned the social-media incentives critique into a public movement and now applies the same framework to AI competition, safety, and institutional behavior.
AI TEXTBOOK, OFF SWITCH INCLUDED
Russell combines a foundational AI textbook, Berkeley research leadership, and decades of work on human-compatible AI and autonomous-weapons risk.
ONE USEFUL THING
Mollick has become one of the most visible interpreters of practical generative AI for workers, educators, managers, and organizations trying to decide what to do with it.
PELICANS ON BIKES
Willison's constant public testing and documentation of new models makes him an unusually trusted interpreter of what AI systems can actually do for developers.
AI BOSSES, NO SHORT ANSWERS
Patel's long-form interviews have become an oral history of the frontier era, giving researchers and executives a venue to explain ideas that later shape industry debate.
AI APPS, NO TEAM REQUIRED
Levels shows how far a single operator can push AI-assisted product development, turning rapid experiments into a visible counterexample to large-team software economics.
IPHONE HACKER VS. THE WORLD
Hotz moves between autonomous driving, open-source machine learning, hardware, and tiny teams, giving him outsized influence among builders who distrust institutional scale.
AI ENGINEER BECAME A JOB TITLE
Shawn Wang helped give the AI Engineer role a name, publication, and conference circuit, shaping how a fast-growing developer class talks about the new stack.
DEEP LEARNING WITHOUT THE IVORY TOWER
Howard's fast.ai work has spent years lowering the barrier between advanced machine learning research and people who want to use it without joining a giant lab.
TWITTER, BITCOIN, DECENTRALIZE EVERYTHING
Dorsey no longer controls Twitter, but Block, capital, and his persistent bets on open protocols keep him relevant to arguments about who should own internet-scale intelligence and infrastructure.
SAPIENS MEETS THE MACHINES
Harari brings AI into a much larger story about information, institutions, and human power, giving him a global audience for arguments about what machine intelligence changes.
SUPERINTELLIGENCE MADE DOOM RESPECTABLE
Bostrom helped make superintelligence and existential risk serious academic subjects, and his frameworks still shape how researchers and institutions describe long-run AI danger.
2045 OR BUST
Kurzweil spent decades popularizing the idea that accelerating computation would converge with human intelligence, and his forecasts still frame public expectations around the Singularity.
THREE HOURS LATER...
Fridman's interviews give frontier researchers, founders, and technology figures unusually long access to a global audience, making his platform a durable agenda-setting venue.
NEOVIM VS. THE ROBOTS
ThePrimeagen has built a large developer audience around programming culture and increasingly AI-assisted coding, giving him real influence over how working engineers receive new tools.
NONZERO-SUM WITH ROBOTS
Wright's long-running work on cooperation, evolution, and moral progress now extends into AI, giving him a smaller but durable role in the intellectual argument over what advanced systems mean.