INHUMAIN.AI
The Watchdog Platform for Inhuman Intelligence
Documenting What Happens When Intelligence Stops Being Human
AI Incidents (2026): 847 ▲ +23% | Countries with AI Laws: 41 ▲ +8 YTD | HUMAIN Partnerships: $23B ▲ +$3B | EU AI Act Fines: €14M ▲ New | AI Safety Funding: $2.1B ▲ +45% | OpenAI Valuation: $157B ▲ +34% | AI Job Displacement: 14M ▲ +2.1M | HUMAIN Watch: ACTIVE 24/7 | AI Incidents (2026): 847 ▲ +23% | Countries with AI Laws: 41 ▲ +8 YTD | HUMAIN Partnerships: $23B ▲ +$3B | EU AI Act Fines: €14M ▲ New | AI Safety Funding: $2.1B ▲ +45% | OpenAI Valuation: $157B ▲ +34% | AI Job Displacement: 14M ▲ +2.1M | HUMAIN Watch: ACTIVE 24/7 |

AI Statistics 2026: Every Number That Matters

The most comprehensive collection of AI statistics for 2026. Market size, investment flows, compute infrastructure, workforce impact, safety metrics, regulatory progress, and research output — with sources and context.

Numbers do not lie, but they can be made to obscure. The AI industry is awash in statistics designed to inflate valuations, justify investments, and manufacture inevitability. This page exists to provide a curated, contextualized reference of the figures that actually matter — with sources, caveats, and the context that press releases leave out.

We update this page monthly. All figures are sourced from public filings, government data, peer-reviewed research, and credible industry analysis. Where estimates diverge significantly across sources, we note the range and explain the discrepancy.

For related analysis, see our AI Regulation Tracker, HUMAIN Tracker, and AI Incident Tracker.


Market Size & Investment

Global AI Market

Metric Value Source
Global AI market size (2025) $244 billion IDC
Projected global AI market (2026) $298 billion IDC
Projected global AI market (2030) $827 billion Grand View Research
Year-over-year growth rate 22.1% Statista
Generative AI market (2025) $67 billion Bloomberg Intelligence
Generative AI projected (2028) $185 billion Bloomberg Intelligence
Enterprise AI spending (2025) $166 billion Gartner
AI share of total IT spending 8.4% Gartner

Venture Capital & Private Investment

Metric Value Year
Total AI VC funding $97 billion 2025
Total AI VC funding $72 billion 2024
Total AI VC funding $49 billion 2023
Number of AI funding rounds 4,200+ 2025
Median AI Series A $18 million 2025
Median AI Series B $52 million 2025
AI share of total VC funding 38% 2025
AI unicorns created 47 2025

Top AI Funding Rounds (2025)

Company Amount Round Lead Investors
OpenAI $6.6 billion Series E Thrive Capital, Microsoft, SoftBank
Anthropic $4.0 billion Series D Amazon, Google, Salesforce
xAI $6.0 billion Series C Valor Equity, Sequoia
Databricks $3.5 billion Late-stage Thrive Capital
CoreWeave $2.0 billion Debt + equity Magnetar Capital
Mistral AI $1.1 billion Series C General Catalyst
Perplexity AI $900 million Series C IVP, Institutional Venture Partners
Figure AI $750 million Series B Microsoft, NVIDIA, Intel

Sovereign & Government AI Investment

Country/Fund AI Investment Commitment Timeframe
United States (federal) $32 billion announced FY2025-2027
Saudi Arabia (HUMAIN + PIF) $100 billion announced 2025-2030
UAE (MGX + ADIA) $30 billion 2025-2028
China (national + provincial) $52 billion estimated 2024-2027
European Union (AI package) $22 billion 2025-2027
United Kingdom $8.5 billion 2025-2030
India (IndiaAI Mission) $1.2 billion 2024-2029
South Korea $7 billion 2025-2029
Japan $6.8 billion 2024-2028
France $3.5 billion 2025-2030
Canada (Pan-Canadian AI Strategy) $1.8 billion (CAD 2.4B) 2024-2029

Context: The HUMAIN announcement of $100 billion represents the single largest sovereign AI commitment by a wide margin, though it includes infrastructure (data centers, energy) alongside pure AI research and development. Whether this capital is deployed effectively — and to whose benefit — is a central question of this publication.


Compute & Infrastructure

GPU Market

Metric Value
NVIDIA data center GPU revenue (2025) $115 billion estimated
NVIDIA AI GPU market share 85-90%
H100 GPUs shipped (2024) ~2.5 million
B200 GPUs shipping (2025) Ramping production
Average H100 price (cloud rental, per hour) $2.50-3.50
Average H100 purchase price $30,000-40,000
Google TPU v5p pods deployed 8,960 chips per pod
AMD MI300X revenue (2025) $9 billion estimated

Training Costs

Model Estimated Training Cost Year
GPT-4 $78 million 2023
Gemini Ultra $100+ million 2023
GPT-4.5 / GPT-5 generation $200-500 million (est.) 2025
Claude 3.5 Sonnet $50-80 million (est.) 2024
Llama 3 405B $60 million (est.) 2024
Frontier model training (2026 est.) $500M-1B 2026

Data Center Infrastructure

Metric Value
AI data center capacity under construction (US) 18+ GW
New AI data center investment (global, 2025) $190 billion
Microsoft AI data center spending (FY2025) $80 billion
Google AI data center spending (2025) $75 billion
Amazon/AWS AI infrastructure (2025) $100 billion
Meta AI infrastructure spending (2025) $60 billion
Number of hyperscale data centers (global) 1,000+
AI share of total data center capacity 34% and growing

Energy & Environmental Impact

Metric Value
AI data center electricity consumption (global, 2025) 134 TWh estimated
Projected AI data center electricity (2028) 325 TWh
AI share of global electricity demand 1.5% (2025), projected 3.5% (2028)
Water consumption per ChatGPT query ~500ml (cooling)
Carbon footprint of training GPT-4 ~5,000 tonnes CO2e (est.)
Google total data center water use (2024) 6.1 billion gallons
Microsoft total data center water use (2024) 2.1 billion gallons

Context: These environmental figures are among the most underreported aspects of the AI boom. Every query, every training run, every model has a physical footprint. For further analysis, see our AI Statistics environmental section updates.


AI Adoption

Enterprise Adoption

Sector AI Adoption Rate (2025) Primary Use Case
Financial services 72% Fraud detection, risk modeling
Healthcare 58% Medical imaging, drug discovery
Manufacturing 55% Predictive maintenance, quality control
Retail/e-commerce 64% Recommendation, pricing, customer service
Technology 83% Code generation, testing, infrastructure
Legal 42% Document review, contract analysis
Education 38% Tutoring, assessment, content creation
Agriculture 28% Crop monitoring, yield prediction
Government 34% Citizen services, fraud detection
Energy 47% Grid optimization, predictive maintenance

Consumer AI Usage

Metric Value
ChatGPT weekly active users 300 million+
Claude monthly active users 50 million+ (est.)
Gemini monthly active users 120 million+ (est.)
Adults who have used generative AI (US) 68%
Adults who use AI tools weekly (US) 34%
Adults who use AI tools daily (US) 12%
Generative AI awareness (global) 87%
Trust in AI-generated content 31%

Developer Tools Adoption

Tool Estimated Users Category
GitHub Copilot 1.8 million paid subscribers Code completion
Cursor 500K+ users AI-native IDE
ChatGPT (code use) 40M+ developers Multi-purpose
Replit AI 25 million registered Code generation
Amazon CodeWhisperer 1 million+ Code completion
Tabnine 1 million+ Code completion

See our AI Tools Database for comprehensive tool listings and reviews.


Workforce Impact

AI Job Market

Metric Value Region
AI-related job postings (2025) 420,000+ United States
AI job postings growth (YoY) +31% United States
AI engineer median salary $185,000 United States
ML engineer median salary $165,000 United States
AI research scientist salary (top labs) $300K-700K+ United States
AI safety researcher salary $180K-400K United States
AI talent shortage (unfilled positions) 1.2 million Global

AI Compensation at Frontier Labs

Role Compensation Range (Total Comp) Notes
Research Scientist (OpenAI) $400K-900K Includes equity
Research Scientist (Anthropic) $350K-700K Includes equity
Research Scientist (DeepMind) $300K-650K Google equity
Senior ML Engineer (frontier lab) $350K-600K
AI Safety Researcher $180K-400K Varies widely
AI Policy Researcher $100K-200K Government roles lower

Workforce Displacement Estimates

Source Prediction Timeframe
McKinsey Global Institute 12 million occupational transitions (US) By 2030
World Economic Forum 85 million jobs displaced, 97 million created By 2030
Goldman Sachs 300 million jobs affected globally By 2030
OECD 27% of jobs at high risk of automation Current
IMF 40% of global employment exposed to AI Current
Brookings Institution 36 million Americans in high-exposure jobs Current

Context: “Affected” and “displaced” are different things. These projections vary dramatically in methodology and definition. Most serious analyses conclude that AI will transform far more jobs than it eliminates entirely — but the transformation itself can be deeply disruptive.


Safety & Incidents

AI Safety Metrics

Metric Value
Documented AI incidents (cumulative) 847 (INHUMAIN.AI tracker)
AI incidents logged (2025) 312
AI incidents rated “critical” (2025) 23
Countries affected by AI incidents 67
AI safety research papers (2025) 2,400+
AI safety research funding (2025) $820 million (est.)
Safety funding as % of total AI investment ~0.8%
Companies with published safety policies 43 (of top 100 AI companies)

For detailed incident data, see our AI Incident Tracker.

AI Safety Funding Sources

Source Amount (2025) Focus
Open Philanthropy $180 million Alignment research, governance
UK AI Safety Institute $120 million Evaluation, testing
US AISI (NIST) $90 million Standards, evaluation
Anthropic (internal safety) $80 million (est.) Constitutional AI, interpretability
Google DeepMind (safety) $100 million (est.) Alignment, governance
NSF AI safety grants $45 million Academic research
EU AI Office $30 million Regulatory capacity
Private foundations (other) $175 million Various

Regulation

Global Regulatory Status

Status Count Examples
Binding AI-specific laws enacted 14 EU, China, Brazil, South Korea, Canada
Executive orders / national strategies 30+ US, India, Japan, UAE, Saudi Arabia
AI regulatory frameworks (non-binding) 40+ OECD, UNESCO, Singapore, Australia
Countries with AI safety institutes 6 UK, US, Japan, Singapore, Canada, EU
International AI governance forums 8 UN AI Advisory Body, GPAI, AI Safety Summit series

Enforcement Actions

Jurisdiction Enforcement Actions (2025) Largest Fine
EU (GDPR, AI-related) 23 $14.5 million
EU (AI Act, prohibited practices) 4 Pending
Italy (Garante) 7 $15.6 million (OpenAI)
China (CAC) 12 Undisclosed
US (FTC) 8 $5.8 million (Rite Aid facial recognition)
South Korea (PIPC) 5 $4.2 million

For detailed regulatory tracking, see our AI Regulation Tracker and EU AI Act enforcement guide.


Research & Development

AI Research Output

Metric Value (2025) Change (YoY)
AI research papers published 185,000+ +18%
AI papers on arXiv 92,000+ +22%
AI patent applications (global) 145,000+ +25%
AI patents granted (US) 38,000+ +30%
Countries producing AI research 120+ Stable
AI PhD graduates (US) 2,800+ +8%

Research by Country (Share of Top AI Papers)

Country Share of Top-Cited AI Papers Trend
United States 32% Stable
China 28% Up
United Kingdom 8% Stable
Germany 4% Stable
Canada 4% Stable
France 3% Up
South Korea 3% Up
Israel 2.5% Stable
Japan 2% Down
India 2% Up

Model Releases (2025)

Model Organization Parameters Open/Closed
GPT-5 OpenAI Undisclosed Closed
Gemini 2.0 Ultra Google DeepMind Undisclosed Closed
Claude 4 (Opus) Anthropic Undisclosed Closed
Llama 4 Meta 400B+ Open weights
Mistral Large 3 Mistral AI Undisclosed Open weights
Grok-3 xAI Undisclosed Partially open
DeepSeek-V3 DeepSeek 685B MoE Open weights
Qwen 3 Alibaba 110B Open weights
Command R+ 2 Cohere Undisclosed Closed
Phi-4 Microsoft 14B Open weights

Benchmark Saturation

Benchmark Best Score (2023) Best Score (2025) Human Baseline
MMLU 86.4% 92.3% ~89.8%
HumanEval (code) 85.0% 95.1% ~95%
GSM8K (math) 92.0% 97.8% ~95%
ARC-Challenge 96.3% 98.7% ~85%
HellaSwag 95.3% 98.1% ~95.6%
GPQA Diamond 41.3% 65.2% ~65% (experts)

Context: Benchmark saturation is a growing problem. When models approach or exceed human baselines on established benchmarks, those benchmarks lose their ability to differentiate capability levels. The field is shifting toward more challenging evaluations (GPQA, SWE-bench, frontier math) and real-world performance metrics.


Key Takeaways

The numbers paint a consistent picture: investment is accelerating, capabilities are advancing, adoption is broadening, and safety infrastructure is not keeping pace. The ratio of safety funding to total AI investment — roughly 0.8% — is perhaps the single most important statistic on this page. It reflects a civilizational bet that the technology being built will not require the safety measures that its own creators acknowledge are necessary.

For context on what these numbers mean for the future, see our AI Doomsday Clock and AI Prediction Scorecard.


This page is updated monthly by the INHUMAIN.AI data team. All figures are best estimates based on available public data. Where proprietary or classified data would provide more accurate figures, we note the limitation. Corrections and updated data points can be submitted through our contact page.