Skip to content

Benchmark

Stanford AI Index (2026)

Stanford's annual report tracking global AI progress, including model performance, investment, policy, and talent trends across countries.

Known aliases

  • 2025 AI Index
  • 2026 AI Index
  • 2026 Index
  • AI Index
  • AI Index 2025
  • AI Index 2026
  • AI Index report
  • Stanford 2026 AI Index
  • Stanford AI Index
  • Stanford AI Index 2026

Current stories

investOne report1 publisher

Senior roles take 71% of the rebound in US software job postings

Senior roles drove 71% of the rise in US software job postings in the year to May 2026, Indeed Hiring Lab data show. Postings are still 27.5% below pre-pandemic levels, and the new openings favor experienced people who can check AI-written code over juniors.

Publishers:cnbc.com

Reality

Evidence58
Adoption55
Hype gap+15
Incentives40
Confidence55
investOne report1 publisher

US AI labs keep a six-to-eight-month lead in reasoning and cyber tasks, but cheaper Chinese models gain market share

Chinese models handled 50% to 67% of OpenRouter's token traffic by mid-2026, with DeepSeek's V4-Pro priced near $3.96 per million output tokens. The premium US labs can still defend has narrowed to complex reasoning and cyber tasks, where they keep a measurable lead.

Reality

Evidence35
Adoption50
Hype gap+25
Incentives
Not scored
Confidence35
leadershipOne report1 publisher

Flagship output tokens now run about what GPT-3.5-level capability cost in 2022

Kion's chief executive argues in Forbes that tokens need cloud-style FinOps discipline, citing a 286-fold fall in per-token inference cost alongside LLM spending that tripled in 2025 and 93 percent of surveyed firms overrunning their AI budgets.

Publishers:forbes.com

Reality

Evidence30
Adoption
Not scored
Hype gap+45
Incentives85
Confidence62
leadershipOne report1 publisher

Before hiring, sort the role's ten tasks into four buckets to decide whether to hire, automate or redesign

An Entrepreneur contributor column asks managers to list the ten recurring tasks behind a requisition and sort them into judgment, relationship, repetition and coordination work. Its supporting studies measured support agents and consultants.

Publishers:entrepreneur.com

Reality

Evidence33
Adoption28
Hype gap+22
Incentives38
Confidence55
productOne report1 publisher

Stanford finds 47% of students unsure whether they are allowed to use AI

Stanford's education chapter measures how many American students are guessing at their school's AI rules. The written policy is the thing product teams have to design around.

Publishers:thenextweb.com

Reality

Evidence60
Adoption78
Hype gap+8
Incentives35
Confidence64
leadershipOne report1 publisher

A Fed note reports AI adoption at 18 percent, 78 percent and 41 percent of the same economy

The three figures count firms, employment and individual workers, and the Census Bureau declared its own earlier adoption numbers incomparable after one clause in the question changed. Boards borrow such numbers to set targets.

Publishers:forbes.com

Reality

Evidence55
Adoption58
Hype gap+10
Incentives60
Confidence52
buildOne report1 publisher

Lightcast prices the AI wage premium at less than half of PwC's 62%

AI skills appear in 2.5% of US job postings, roughly a million of them, and the two most-cited estimates of what they add to pay within the same occupation differ by more than two to one, with only one of them expressed in dollars.

Publishers:dev.to

Reality

Evidence40
Adoption55
Hype gap+28
Incentives62
Confidence45
investOne report1 publisher

Huang rejects circular financing charge, says NVIDIA's customer stakes are small relative to demand

Cryptopolitan reports that the NVIDIA chief told Goldman's conference his stakes in OpenAI and CoreWeave are small next to the demand they create, and the two disclosed equity cheques do come to only $32bn.

Reality

Evidence38
Adoption61
Hype gap+24
Incentives71
Confidence37
investOne report1 publisher

a16z lifts its growth fund to $8.5bn days after committing $1.1bn to AI hardware

Two vehicles raised days apart put $9.6bn behind AI, and the smaller one buys into a layer where a16z says rack power runs from 250 kilowatts today to a megawatt within three years. Supply, on the firm's own numbers, grows 20-30% a year.

Reality

Evidence60
Adoption25
Hype gap+12
Incentives74
Confidence55
investOne report1 publisher

SEC presses advisers to prove their OpenAI and Anthropic SPVs hold real shares

A reported examination sweep has landed on a market where two issuers alone carry roughly $2.28 trillion of estimated paper value. Both issuers say they never authorised the vehicles selling access to their shares.

Reality

Evidence41
Adoption48
Hype gap+31
Incentives61
Confidence37
investOne report1 publisher

Anthropic and OpenAI added $1.74 trillion of private paper value in twelve months

DeFiLlama now marks the two labs at $2.28 trillion against an implied $544 billion a year ago, which leaves every bank pitching an AI listing to price against a reference no public order book has tested.

Reality

Evidence28
Adoption
Not scored
Hype gap+55
Incentives72
Confidence36
leadershipOne report1 publisher

Re-baseline AI procurement on cost per completed task, not dollars per million tokens

Per-token prices fell about 200x since GPT-4's launch while US enterprise AI spend tripled to $37 billion. Tokenizer variance, reasoning tokens and tier discounts are where the bill diverges.

Publishers:hexaware.com

Reality

Evidence55
Adoption58
Hype gap+15
Incentives68
Confidence45
investOne report1 publisher

Korea's 720-billion-dollar AI plan meets its first critic: the man running its research hub

Seoul wants top-three status by 2035 on more than 1,000 trillion won. The inaugural director of its National AI Research Hub says the money is budgeted on the wrong clock.

Publishers:en.sedaily.com

Reality

Evidence32
Adoption28
Hype gap+22
Incentives66
Confidence38