AI Market Growth Statistics Updated November 2026

Human Written & Fact Checked · Updated

100 AI Market Growth Statistics for 2026

Which AI market are you measuring? A forecast of worldwide spending, a chipmaker's sales, cloud revenue and a survey of business adoption can all rise together without measuring the same thing. Adding them would double-count activity.

This hub keeps those measures separate. Forecasts retain their publication vintage, company disclosures retain their reporting period, and annual adoption surveys remain historical snapshots. Use the distinctions to evaluate growth claims before comparing their headline numbers.

Headline figures

  • $2.670460 trillion: Forecast total AI spending (Worldwide; 2026 forecast). (Gartner, Sep 2026)
  • 49.5%: Forecast annual AI spending growth (Worldwide; 2026 forecast). (Gartner, Sep 2026)
  • $121.4 billion: AI infrastructure spending (Worldwide; Q2 2026). (IDC, Nov 2026)
  • 53.3%: Annual infrastructure spending growth (Worldwide; Q2 2026). (IDC, Nov 2026)
  • $92.8 billion: United States AI infrastructure spending (Regional; Q2 2026). (IDC, Nov 2026)
  • $89.0 billion: NVIDIA Data Center revenue (Worldwide; Quarter ended July 26, 2026). (NVIDIA, Aug 2026)
  • More than 1 billion: OpenAI product weekly active users (Worldwide; September 2026 disclosure). (OpenAI, Sep 2026)
  • 2.5 million: Businesses reached by OpenAI products (Worldwide; September 2026 disclosure). (OpenAI, Sep 2026)
  • 20.0%: Eligible EU enterprises using AI (European Union; 2025). (Eurostat, Dec 2025)
  • 55.03%: Large-enterprise AI adoption (European Union; 2025). (Eurostat, Dec 2025)
  • Nearly 17x: Bedrock token volume versus prior year (Worldwide; Q3 2026 disclosure). (Amazon, Oct 2026)
  • Nearly 32 billion: First-party model API tokens processed per minute (Worldwide; Q3 2026 disclosure). (Alphabet, Oct 2026)

Browse AI market statistics

What does the worldwide AI spending forecast cover?

The category definition counts as much as the total. Gartner's infrastructure-inclusive forecast is broader than a software-only market estimate.

What was forecast for the current calendar year?

  • $576.481 billion: Forecast AI services spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $51.347 billion: Forecast AI cybersecurity spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $461.637 billion: Forecast AI software spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $29.219 billion: Forecast AI agent and assistant spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $28.266 billion: Forecast generative-model spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $26.445 billion: Forecast data-science and machine-learning platform spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $9.541 billion: Forecast AI application-development platform spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $3.126 billion: Forecast AI data spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)
  • $1.484397 trillion: Forecast AI infrastructure spending. Worldwide; 2026 forecast. (Gartner, Sep 2026)

Infrastructure was the largest spending category in Gartner's September forecast. Every value below is an expected calendar-year amount, not audited vendor revenue or a final market outcome.

How does the following-year forecast compare?

  • $3.637292 trillion: Forecast total AI spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $745.655 billion: Forecast AI services spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $85.997 billion: Forecast AI cybersecurity spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $656.353 billion: Forecast AI software spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $65.472 billion: Forecast AI agent/assistant spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $51.620 billion: Forecast generative-model spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $35.552 billion: Forecast data-science/machine-learning platform spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $12.478 billion: Forecast AI application-development platform spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $6.480 billion: Forecast AI data spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)
  • $1.977685 trillion: Forecast AI infrastructure spending. Worldwide; 2027 forecast. (Gartner, Sep 2026)

Gartner forecast expansion across the categories shown for the following year. The unchanged September vintage makes this a coherent comparison, but it remains a projection that later forecasts can revise.

How much AI infrastructure spending was observed?

Actual infrastructure tracking provides a separate check on the investment story. Price increases can raise spending even when physical capacity grows more slowly.

What did infrastructure buyers spend?

  • 34%: Sequential infrastructure spending growth. Worldwide; Q2 2026. (IDC, Nov 2026)
  • $119.0 billion: AI server spending. Worldwide; Q2 2026. (IDC, Nov 2026)
  • $2.5 billion: AI storage spending. Worldwide; Q2 2026. (IDC, Nov 2026)

IDC recorded growing infrastructure spending, concentrated heavily in servers. The tracker's scope differs from Gartner's broader infrastructure category, so the two totals cannot be added or treated as contradictory measurements. The market value measures what buyers spent under the tracker's definitions. It does not tell you how much usable capacity came online during the period or how heavily that capacity was utilized. For a demand assessment, keep spending, deliveries, installation and actual workloads as separate measures. A strong purchasing quarter can precede the revenue generated by the infrastructure it helps build.

How did accelerated and non-accelerated systems compare?

  • $105.8 billion: Accelerated AI infrastructure spending. Worldwide; Q2 2026. (IDC, Nov 2026)
  • $15.6 billion: Non-accelerated AI infrastructure spending. Worldwide; Q2 2026. (IDC, Nov 2026)
  • 52.2%: Accelerated infrastructure spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)
  • 61.1%: Non-accelerated infrastructure spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)

Accelerated systems accounted for the larger spending amount, while non-accelerated systems grew faster year over year. Both support AI workloads; this comparison measures spending, not accelerator units or compute capacity. The comparison concerns two hardware categories serving a wider computing system. It should not be turned into a claim that one architecture is replacing the other. For interpreting market growth, the relevant question is which workloads and pricing changes explain the spending mix. This published extract does not provide a complete workload-by-workload decomposition.

Where is AI infrastructure investment growing?

A large growth percentage can come from a small base. Regional spending levels and annual growth need to be read together.

Which regions accounted for infrastructure spending?

  • $13.0 billion: China AI infrastructure spending. Regional; Q2 2026. (IDC, Nov 2026)
  • $6.4 billion: Western Europe AI infrastructure spending. Regional; Q2 2026. (IDC, Nov 2026)
  • $5.9 billion: Asia Pacific excluding Japan and China infrastructure spending. Regional; Q2 2026. (IDC, Nov 2026)
  • $0.9 billion: Middle East and Africa AI infrastructure spending. Regional; Q2 2026. (IDC, Nov 2026)
  • $0.3 billion: Central and Eastern Europe AI infrastructure spending. Regional; Q2 2026. (IDC, Nov 2026)

The United States accounted for the largest regional infrastructure spending amount reported here. IDC's rounded geographic categories include a separate Asia Pacific group excluding Japan and China. Regional totals describe the tracker's market allocation. They do not identify every end user served by the equipment or imply that the benefits remain within the purchasing region. Keep that distinction when comparing investment with national adoption surveys.

Which regions grew fastest?

  • 58.7%: United States infrastructure spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)
  • 27.9%: China infrastructure spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)
  • 83.9%: Western Europe infrastructure spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)
  • 28.5%: Asia Pacific excluding Japan/China spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)
  • 108.9%: Middle East/Africa infrastructure spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)
  • 96.7%: Central/Eastern Europe infrastructure spending growth. Regional and worldwide; Q2 2026 year-on-year. (IDC, Nov 2026)

Growth was uneven across regions, with smaller markets sometimes posting faster percentage gains. These are year-on-year changes in spending; they do not directly measure market-share gains, installed capacity or workload growth.

What do suppliers and AI platforms disclose?

Company metrics expose demand, but definitions still differ. Cloud segments contain non-AI workloads, user counts are not paid subscriptions, and tokens are not revenue.

What do infrastructure suppliers report?

  • 117%: NVIDIA Data Center annual revenue growth. Worldwide; Quarter ended July 26, 2026. (NVIDIA, Aug 2026)
  • 18%: NVIDIA Data Center sequential revenue growth. Worldwide; Quarter ended July 26, 2026. (NVIDIA, Aug 2026)
  • $63.9 billion: Microsoft Cloud revenue. Worldwide; Q3 2026. (Microsoft, Oct 2026)
  • 30%: Microsoft Cloud annual revenue growth. Worldwide; Q3 2026. (Microsoft, Oct 2026)
  • 48%: Azure annual revenue growth. Worldwide; Q3 2026. (Microsoft, Oct 2026)
  • $30.7 billion: Google Cloud revenue. Worldwide; Q3 2026 disclosure. (Alphabet, Oct 2026)
  • 103%: Google Cloud annual revenue growth. Worldwide; Q3 2026 disclosure. (Alphabet, Oct 2026)

NVIDIA's Data Center segment and the major cloud businesses reported growth. Their published segments include more than narrowly defined AI products, so these figures are infrastructure context and cannot be used as pure-AI market revenue.

How widely are AI services being used?

  • Nearly 7x: Fortune500 Bedrock spending versus prior year. Worldwide; Q3 2026 disclosure. (Amazon, Oct 2026)
  • 2x: AgentCore agents deployed versus preceding three quarters combined. Worldwide; Q3 2026 disclosure. (Amazon, Oct 2026)
  • 400 million+: Alexa Shopping customers during trailing twelve months. Worldwide; Q3 2026 disclosure. (Amazon, Oct 2026)
  • 570%: Alexa Shopping annual interaction growth. Worldwide; Q3 2026 disclosure. (Amazon, Oct 2026)
  • 1 billion: Gemini app users reported. Worldwide; Q3 2026 disclosure. (Alphabet, Oct 2026)

The disclosed platforms show large audiences and expanding activity, but their metrics are not comparable user populations. Weekly users, businesses reached, app users, token volume and agent deployments each answer a distinctly different demand question. Treat each disclosure as evidence of its specific commercial activity, not interchangeable market penetration.

What capital and capacity commitments were announced?

  • $65 billion: Anthropic Series H capital raised. Worldwide; May 2026 announcement. (Anthropic, May 2026)
  • $965 billion: Anthropic post-money financing valuation. Worldwide; May 2026 announcement. (Anthropic, May 2026)
  • Above $47 billion: Anthropic annualized revenue run rate. Worldwide; May 2026 announcement. (Anthropic, May 2026)
  • Up to 5 GW: Contracted new Anthropic capacity with Amazon. Worldwide; May 2026 announcement. (Anthropic, May 2026)
  • 5 GW: Contracted Anthropic TPU capacity with Google and Broadcom. Worldwide; May 2026 announcement. (Anthropic, May 2026)

Anthropic's May disclosure described announced financing, annualized revenue and compute commitments. It is a historical snapshot: financing valuation is not revenue, and contracted gigawatts do not establish that capacity was already installed. Assess announced commitments separately from delivery schedules, operating utilization and subsequently recognized revenue.

How broadly have businesses adopted AI?

Adoption surveys help distinguish a growing supplier market from widespread operational use. Their coverage and reference periods differ, so cross-country comparisons need care.

How did adoption vary across Europe?

  • 42.0%: Eligible Danish enterprises using AI. European Union; 2025. (Eurostat, Dec 2025)
  • 37.8%: Eligible Finnish enterprises using AI. European Union; 2025. (Eurostat, Dec 2025)
  • 35.0%: Eligible Swedish enterprises using AI. European Union; 2025. (Eurostat, Dec 2025)
  • 5.2%: Eligible Romanian enterprises using AI. European Union; 2025. (Eurostat, Dec 2025)
  • 8.4%: Eligible Polish enterprises using AI. European Union; 2025. (Eurostat, Dec 2025)
  • 8.5%: Eligible Bulgarian enterprises using AI, initial release. European Union; 2025. (Eurostat, Dec 2025)
  • 17%: Small-enterprise AI adoption. European Union; 2025. (Eurostat, Dec 2025)
  • 30.36%: Medium-enterprise AI adoption. European Union; 2025. (Eurostat, Dec 2025)
  • 62.52%: Information/communication enterprise AI adoption. European Union; 2025. (Eurostat, Dec 2025)
  • 40.43%: Professional/scientific/technical enterprise AI adoption. European Union; 2025. (Eurostat, Dec 2025)
  • 24.82%: Real-estate enterprise AI adoption. European Union; 2025. (Eurostat, Dec 2025)
  • 10.79%: Construction enterprise AI adoption. European Union; 2025. (Eurostat, Dec 2025)

Adoption varied by country, size and activity in Eurostat's annual survey. Its employment threshold excludes microbusinesses; covered sectors do not represent the whole economy.

Which technologies were European enterprises using?

  • 11.8%: Eligible EU enterprises using text-analysis AI. European Union; 2025. (Eurostat, Dec 2025)
  • 9.5%: Eligible EU enterprises generating images, video or audio. European Union; 2025. (Eurostat, Dec 2025)
  • 8.8%: Eligible EU enterprises generating written or spoken language. European Union; 2025. (Eurostat, Dec 2025)
  • 7.2%: Eligible EU enterprises using speech-recognition AI. European Union; 2025. (Eurostat, Dec 2025)
  • 1.39%: Enterprises using autonomous-movement AI. European Union; 2025. (Eurostat, Dec 2025)
  • 35.04%: Large enterprises using text-mining AI. European Union; 2025. (Eurostat, Dec 2025)
  • 31.68%: Large enterprises using language-generation AI. European Union; 2025. (Eurostat, Dec 2025)
  • 8.55%: Large enterprises using autonomous-movement AI. European Union; 2025. (Eurostat, Dec 2025)

Text and language applications were reported more commonly than autonomous physical movement. The all-enterprise and large-enterprise figures have different denominators, and a respondent can use several technologies at once.

What did the US business survey show?

  • 19.8%: Employer businesses using AI in any business function. US; May 2026. (US Census Bureau, May 2026)
  • 32%: Businesses with 100–249 employees using AI. US; May 2026. (US Census Bureau, May 2026)
  • 39.7%: Information-sector businesses using AI. US; May 2026. (US Census Bureau, May 2026)
  • 33.9%: Finance and insurance businesses using AI. US; May 2026. (US Census Bureau, May 2026)
  • About 14%: Retail businesses using AI. US; May 2026. (US Census Bureau, May 2026)
  • About 42%: Information businesses expecting AI use within six months. US; May 2026. (US Census Bureau, May 2026)
  • About 39%: Finance and insurance businesses expecting AI use. US; May 2026. (US Census Bureau, May 2026)
  • About 17%: Retail businesses expecting AI use within six months. US; May 2026. (US Census Bureau, May 2026)

In the historical fortnight ending 3 May 2026, information and finance businesses reported higher use than the overall employer-business rate, while retail was lower. Expected use is a separate question; these figures do not claim current November prevalence.

What determines whether adoption becomes useful work?

A reported deployment is only the beginning of the commercial story. Task selection and barriers help explain why a market can grow faster than adoption spreads.

What tasks did AI-using enterprises support?

  • 34.70%: AI-using enterprises applying AI to marketing/sales. European Union; 2025. (Eurostat, Dec 2025)
  • 31.05%: AI-using enterprises applying AI to administration. European Union; 2025. (Eurostat, Dec 2025)
  • 6.08%: AI-using enterprises applying AI to logistics. European Union; 2025. (Eurostat, Dec 2025)

Among AI adopters, marketing and administration were more common uses than logistics. The denominator is businesses already using AI, so these values cannot be read as shares of all enterprises. A use-case share is not a measure of economic value produced. It also does not show the share of staff, spending or transactions handled by AI. A company might use AI in marketing and administration simultaneously, so adding the categories would count some businesses more than once. The useful next question is how each application changed a defined workflow and whether that change was measured.

What held interested non-adopters back?

  • 70.89%: Interested non-adopters citing expertise gaps. European Union; 2025. (Eurostat, Dec 2025)
  • 52.52%: Interested non-adopters citing legal uncertainty. European Union; 2025. (Eurostat, Dec 2025)
  • 48.83%: Interested non-adopters citing privacy concerns. European Union; 2025. (Eurostat, Dec 2025)

Lack of expertise was the most common barrier among interested non-adopters, ahead of legal uncertainty and privacy concerns. These respondents had considered AI but were not using it; they are a selected population. Interest and deployment are different stages. A business can recognize a potential application yet lack the skills or confidence to implement it. That makes the denominator essential: this is evidence about businesses that considered the technology, not proof that the same obstacles explain every non-user. Distinguish implementation support, internal governance and unresolved legal questions before treating any one barrier as a purchasing requirement.

Glossary

Methodology and sources

Every figure on this page comes from the organisation that produced it: regulators, statistics offices, public bodies, research institutes, and companies reporting their own data or surveys. Each one links to the original report or dataset and carries its release date. Survey results keep their population and fieldwork period, and financial figures keep the measure the producer used. Figures from different producers are not added together or ranked against each other. The page is reviewed monthly and updated when a producer releases newer data.