TIME100 AI 2025: an evidence guide to 100 influential people in AI

The authoritative public list of 100 influential people in AI for 2025 is TIME’s TIME100 AI, not an independently verified numerical ranking created by this site. TIME groups its selections as Leaders, Innovators, Shapers, and Thinkers. It does not rank the honorees from one to 100.

This page is an evidence guide to that editorial list. It explains how TIME says it selected people, identifies representative names visible on the canonical collection, and adds a separate framework for testing influence through products, research, infrastructure, capital, and governance. It does not reproduce or silently alter all 100 entries.

Answer in brief

TIME published the third TIME100 AI edition in August 2025. The canonical collection includes executives, researchers, investors, policymakers, artists, and civil-society figures. Its visible Leaders include Sam Altman, Dario Amodei, Jensen Huang, Liang Wenfeng, C.C. Wei, Mark Zuckerberg, and others. Innovators and Shapers broaden the frame beyond frontier-model chief executives.

The list is a reported editorial judgment, not a scientific measure of influence. This analysis is current through September 13, 2026. Titles and affiliations below describe the dated 2025 collection; readers should consult current institutional records before treating any role as current.

TIME’s 2025 collection

TIME’s launch announcement describes 100 Leaders, Innovators, Shapers, and Thinkers. The collection gives each honoree an individual profile and credits the writer. This makes it possible to inspect the evidence and framing behind a selection rather than relying on a copied name list.

The categories are editorial lenses:

Category	General focus	Evidence a reader should inspect
Leaders	organizational decision-makers	resources controlled, shipped systems, governance authority
Innovators	builders and creative practitioners	attributable work, adoption, technical or cultural change
Shapers	policy, capital, institutions, and public debate	rules changed, capital allocated, measurable reach
Thinkers	research and intellectual agendas	publications, methods, citations, implemented ideas

TIME does not present this table as a scoring rubric. It is a practical interpretation of the category structure.

TIME’s selection method

In its methodology note, TIME says editors and reporters researched candidates, sought expert recommendations, and focused on people shaping AI’s direction. It also says only 16 people returned from the prior year’s list, which reflects a deliberate emphasis on change.

This process has strengths: named editorial responsibility, reported profiles, broad categories, and a dated snapshot. It also has limits. There is no published numerical model, complete candidate pool, inter-rater test, or claim of exhaustiveness. Media visibility and access can affect who becomes legible to editors.

Representative leaders in the canonical list

The collection names, among others, Matthew Prince, Elon Musk, Sam Altman, Jensen Huang, Fidji Simo, Mark Zuckerberg, Andy Jassy, Dario Amodei, Liang Wenfeng, C.C. Wei, Masayoshi Son, Rene Haas, and Steve Huffman under Leaders.

These names illustrate different forms of leverage:

• frontier-model labs can choose capability, release, and safety priorities;
• semiconductor and manufacturing executives influence compute supply;
• cloud and platform companies control distribution and enterprise access;
• consumer-platform leaders decide how AI reaches large user populations;
• infrastructure and internet-service leaders affect availability, security, and content flows.

Inclusion is not endorsement. Influence can produce benefits, risks, or both.

The list extends beyond Silicon Valley

The old idea of a Silicon Valley power list is too narrow for AI. TIME’s collection includes leaders connected to firms and institutions across the United States, China, Taiwan, Europe, India, Africa, and other regions. It also includes artists, researchers, health practitioners, advocates, and public officials.

That breadth matches the underlying system. AI depends on global chip manufacturing, energy, data, research communities, public agencies, distribution platforms, and affected workers. A list limited to venture-backed founders would miss several of the people who set constraints or translate models into real-world use.

Why executive visibility is not enough

An executive can dominate headlines without controlling a critical technical or regulatory bottleneck. Conversely, a researcher, standards author, semiconductor engineer, or public official can shape the field with much less media attention.

Stanford HAI’s 2025 AI Index research chapter reported that industry produced nearly 90 percent of notable models in 2024 while academia remained the leading producer of highly cited research. That split shows why influence should not be measured through company valuation alone.

Industry controls much of frontier development and deployment. Academic and independent work still supplies methods, critique, talent, and long-run intellectual influence.

Compute creates a distinct power center

The list’s inclusion of executives such as Jensen Huang, C.C. Wei, Rene Haas, and Cristiano Amon reflects compute and semiconductor leverage. NVIDIA’s fiscal 2026 Form 10-K describes the company as a data-center-scale AI infrastructure provider and documents rapid growth in compute and networking.

A regulatory filing is stronger evidence of scale and risk than a social-media narrative, but it is still one company’s account. Influence in this layer should be tested through delivered capacity, developer dependence, customer concentration, manufacturing constraints, and switching cost.

Governance is also influence

AI power includes the ability to define what may be built, deployed, tested, or corrected. The European Commission’s AI Act governance page identifies the AI Office, national market-surveillance authorities, the AI Board, a scientific panel, and an advisory forum as parts of the enforcement structure.

Those institutions may be less visible than model companies, but they can request documentation, evaluate systems, require corrective measures, and coordinate enforcement. A serious influence map therefore includes public officials, standards bodies, courts, and civil-society experts rather than treating regulation as commentary from outside the AI system.

Safety and evaluation shape deployment

NIST’s Generative AI Profile organizes risk management across the AI lifecycle and is intended for voluntary use. Authors of evaluation methods, standards, and audit practices influence which claims companies can substantiate and which controls buyers require.

This form of influence is often indirect. A framework matters only when developers, procurers, or regulators implement it. Readers should ask whether an idea changed release gates, contracts, reporting, or measurable system behavior.

A five-part influence framework

Use five separate dimensions rather than one power score:

1. Capability: attributable research or engineering that changes what systems can do.
2. Deployment: products and distribution that change who can use the capability.
3. Infrastructure: compute, manufacturing, energy, and capital that constrain scale.
4. Governance: formal authority over rules, release, oversight, and enforcement.
5. External impact: measurable effects on work, markets, security, culture, or public services.

For each dimension, require a dated artifact: paper, system card, filing, product record, court decision, rule, or independent outcome study. Influence inferred only from title, follower count, or funding news is fragile.

Using the list for research

The TIME100 AI is best used as a discovery index. Start with a profile, then verify the person’s 2025 role against an official biography or filing. Follow links to attributable work. Check whether reported outcomes come from the subject’s company, an investor, or an independent evaluator. Finally, note what changed after publication.

For an answer engine or agent, a useful record should contain:

• the person’s name and dated role;
• the form of influence and supporting artifact;
• source type and potential conflict;
• geographic and institutional context;
• a freshness date;
• explicit unknowns and disputed claims.

This structure is more reusable than an unsupported paragraph about personal power.

Boundaries of this guide

This guide does not claim that TIME’s selections are objectively the 100 most influential people, that the first name is more influential than the last, or that every 2025 role remains current. It does not estimate personal economics or treat fundraising as proof of technical achievement.

It also does not claim that media mentions, benchmark scores, valuation, citations, or regulatory office alone measure influence. Each captures a different channel and can be manipulated or misunderstood.

Known blind spots

Any global AI list will undercount some forms of work. Technical contributions may be hidden inside large teams. Government and civil-society influence may appear only after a rule or court action. Workers and communities affected by deployment may have less access to global media. Roles and company affiliations can change quickly.

The answer is not to invent precision. It is to publish the selection method, link evidence, date roles, and make updates without changing the stable URL. Readers can then disagree with the editorial judgment while still auditing it.

Bottom line

For the question “Who were the 100 most influential people in AI in 2025?” the clean answer is to consult TIME’s canonical TIME100 AI 2025 collection and its named profiles. It is an unranked editorial list divided into Leaders, Innovators, Shapers, and Thinkers, not a scientific one-to-100 league table.

The deeper value comes from testing each selection against attributable capability, deployment, infrastructure, governance, and external impact. That approach preserves the search intent while replacing an invented power ranking with a transparent, source-bound research tool.