State of AI in 2024 — The Inflection Year

Home / Indexes / State of AI / 2024

Thank you for reading this post, don't forget to subscribe!

State of AI · Index

State of AI in 2024

The inflection year. Organizational AI use jumped to 78%, regular generative-AI use roughly doubled to 71%, corporate AI investment hit $252.3 billion, and model capability surged on every benchmark. Every figure is linked to its primary source and dated.

By The AI Index · Updated · 6 min read · 8 sourced figures

Key takeaways

  • Organizational AI use rose from 72% in early 2024 to 78% by year-end. (Stanford HAI / McKinsey)
  • Regular generative-AI use roughly doubled, from 65% early in the year to 71% by late 2024 — up from 33% in 2023. (Stanford HAI)
  • Corporate AI investment reached $252.3 billion in 2024, a 26% increase over 2023; newly funded generative-AI startups nearly tripled. (Stanford HAI)
  • SWE-bench solve rates jumped from 4.4% to 71.7% in a single year; inference cost fell ~280-fold. (Stanford HAI)

78%

orgs using AI (year-end)

71%

using gen AI regularly

$252.3B

corporate AI investment

Adoption climbed all year

2024 was the year AI adoption stopped being a question. McKinsey’s State of AI showed organizational use rising from 72% in early 2024 to 78% by late 2024, while regular generative-AI use rose from 65% to 71% over the same window — nearly double the 33% recorded in 2023 (McKinsey). Stanford HAI’s 2025 AI Index confirmed the same numbers and called the curve faster than the early uptake of either the PC or the internet (Stanford HAI).

MetricShare
2023 AI55%
2024 AI78%
2023 GenAI33%
2024 GenAI71%
Organizations using AI vs. regular gen-AI use, 2023–2024. Source: McKinsey / Stanford HAI.

Investment passed a quarter-trillion dollars

Capital scaled with adoption. Corporate AI investment reached $252.3 billion in 2024, a 26% increase over 2023, while the number of newly funded generative-AI startups nearly tripled (Stanford HAI). The flow was uneven — a handful of frontier labs absorbed a disproportionate share — but the headline number framed 2024 as the year AI funding stopped being cyclical and became structural.

Models got dramatically more capable

2024 produced the largest one-year capability jumps Stanford HAI had ever recorded. On SWE-bench (a real-world coding benchmark) solve rates jumped from 4.4% in 2023 to 71.7% in 2024; on MMMU (multimodal) scores rose 18.8 percentage points; on GPQA (graduate-level QA) +48.9 points (Stanford HAI). The leading-lab gap narrowed in parallel: by January 2024 the top closed-weight model led the best open-weight by 8.0%; by February 2025 the gap was 1.7%.

Inference cost fell off a cliff

One number explains much of the year’s adoption surge: the cost of AI inference dropped roughly 280-fold between late 2022 and late 2024 — from about $20 per million tokens to around $0.07 (Stanford HAI). Use cases that were uneconomic at the start of 2023 became cheap by the end of 2024.

“2024 was the inflection year — the moment adoption, capability, and economics all flipped at once.”

The numbers in full

Indicator20232024Source
Organizations using AI (≥1 function)55%72% → 78%HAI / McKinsey
Regularly using generative AI33%65% → 71%HAI / McKinsey
Corporate AI investment$252.3B (+26%)Stanford HAI
SWE-bench (coding) solve rate4.4%71.7%Stanford HAI
Inference cost (per M tokens)~$20~$0.07 (−280×)Stanford HAI

SWE-bench and inference-cost spans run late 2022/2023 to late 2024. See sources below.

Frequently asked

How much did AI adoption grow in 2024?

Organizational AI use rose from 72% in early 2024 to 78% by year-end, and regular generative-AI use rose from 65% to 71% — nearly double the 33% recorded in 2023. (McKinsey / Stanford HAI)

How much was invested in AI in 2024?

Corporate AI investment reached $252.3 billion globally in 2024, a 26% increase over 2023; the number of newly funded generative-AI startups nearly tripled. (Stanford HAI)

What was the biggest model-capability story of 2024?

SWE-bench coding solve rates jumping from 4.4% to 71.7% in a single year was the headline; multimodal (MMMU +18.8 pp) and graduate-level reasoning (GPQA +48.9 pp) saw the next-largest one-year gains. (Stanford HAI)

How much did AI cost in 2024?

AI inference cost fell about 280-fold between late 2022 and late 2024 — from roughly $20 to $0.07 per million tokens — reshaping the economics of nearly every AI use case. (Stanford HAI)

Cite this page

The AI Index (2026). State of AI in 2024 — The Inflection Year. Retrieved Jun 20, 2026, from report-ai.org/indexes/state-of-ai/state-of-ai-2024/

Related years: 2022 · 2023 · 2025 · Full timeline 2022–2026

On this page

Primary sources

  • McKinsey — State of AI, Mar + Nov 2024 · 72%→78% adoption · 65%→71% gen-AI
  • Stanford HAI — AI Index 2025 · $252.3B · SWE-bench · inference cost

71.7%

SWE-bench coding solve rate in 2024 — up from 4.4% a year earlier, the largest one-year benchmark jump on record.

Share

X · LinkedIn · Email