Research library
The evidence behind local AI
A plain-language guide to published research on AI energy use, public opinion, smaller models, open models, and environmental impact. Every finding links directly to its source.
How to read this page
Start with the large key figure, then read the explanation and any important context. Select the source link to check the original publication. Faded chart marks are estimates.
Topic 1 of 5: Energy
Energy use and grid demand
How much electricity AI and data centres use, from a single prompt to nationwide demand.
415
945
1,200
2024
2030est.
2035est.
About 1.5% of world electricity in 2024. The 2030 figure is slightly more than Japan uses today; across the IEA's other cases, 2035 spans 700 to 1,700 TWh.International Energy Agency, Energy and AI (2025)
Key findings
Key figure
4.4% → 12%
US data centres used 176 TWh in 2023, about 4.4% of national electricity. Berkeley Lab's 2028 range is 325 to 580 TWh, or 6.7% to 12% of everything the country generates.
Source: Lawrence Berkeley National Laboratory, 2024 United States Data Center Energy Usage Report (2024)
Key figure
0.24 Wh
Google's own measurement of a median Gemini Apps text prompt: 0.24 watt-hours of energy, 0.03 gCO₂e, and 0.26 mL of water. Energy and carbon per prompt fell 33× and 44× over the preceding twelve months.
Important context
A per-prompt floor measured by the operator, covering their own serving stack. It is not a total, and it is not independently audited.
Source: Google, Measuring the environmental impact of delivering AI at Google scale (2025)
Key figure
29.6 GW
AI data centre capacity tracked by the 2026 AI Index reached 29.6 GW, comparable to New York's peak demand.
Source: Stanford HAI, Inside the AI Index: 12 Takeaways from the 2026 Report (2026)
Topic 2 of 5: Public sentiment
How people feel about AI
What repeated public surveys show about concern, trust, jobs, and control.
37%
50%
52%
2021
2025
2026
Pew's June waves; the 2025 point is from its June 2025 survey. Pew has asked the same question since 2021.Pew Research Center, Young adults in the US are increasingly wary of AI, concerned it will take jobs (2026)
Key findings
Key figure
55%
Among adults under 30, 55% are now more concerned than excited about AI, the first time a majority of that group has said so.
Important context
Surveyed June 22 to 28, 2026.
Source: Pew Research Center, Young adults in the US are increasingly wary of AI, concerned it will take jobs (2026)
Key figure
71%
71% of US adults think AI will lead to fewer jobs in the United States over the next two decades, up from 64% in 2024.
Source: Pew Research Center, Young adults in the US are increasingly wary of AI, concerned it will take jobs (2026)
Key figure
Half or more
Half or more of both the US public and AI experts say they have little or no control over how AI is used in their lives. More than half of each group want more control.
Source: Pew Research Center, What the data says about Americans' views of artificial intelligence (2026)
Key figure
34% vs 16%
Across 25 countries, a median of 34% are more concerned than excited about AI and 16% are more excited than concerned. Trust to regulate AI runs 53% for the EU, 37% for the US, and 27% for China.
Important context
28,333 adults across 25 countries, surveyed January to April 2025.
Source: Pew Research Center, How People Around the World View AI (2025)
Key figure
31%
US trust in the government to regulate AI sits at 31%, and only 33% of Americans expect AI to improve their own job, against 40% globally.
Source: Stanford HAI, Inside the AI Index: 12 Takeaways from the 2026 Report (2026)
Topic 3 of 5: Local models
Why smaller models matter
Download trends, falling costs, and the growth of models that can run on personal computers.
Qwen
39.6M
Gemma
20.8M
Llama
7.5M
GGUF is the llama.cpp format: these are downloads of models packaged to run on someone's own machine.Hugging Face, State of Open Models: Summer 2026 (2026)
Key findings
Key figure
83%
Models under 1B parameters account for 83% of all-time downloads on the Hugging Face Hub. Models above 100B account for 1%. In 2026 so far, 3% of download volume went to models above 70B.
Source: Hugging Face, State of Open Models: Summer 2026 (2026)
Key figure
+464%
GGUF repositories on the Hub grew 464% this year, and Apple's MLX grew 148%. The local runtime layer is growing faster than the models themselves.
Source: Hugging Face, State of Open Models: Summer 2026 (2026)
Key figure
280×
The inference cost of a system performing at GPT-3.5 level dropped more than 280-fold between November 2022 and October 2024, from roughly $20 to $0.07 per million tokens, driven by increasingly capable small models. Hardware costs fell 30% a year; energy efficiency improved 40% a year.
Source: Stanford HAI, The 2025 AI Index Report (2025)
Key figure
90× fewer
OLMo 3.1 Think 32B, with nearly 90 times fewer parameters than Grok 4, achieves comparable results on several benchmarks.
Important context
Several benchmarks, not all of them. A 32B model still needs more memory than a laptop usually has.
Source: Stanford HAI, The 2026 AI Index Report: Research and Development (2026)
Topic 4 of 5: Open source
How widely open models are used
Adoption, ecosystem size, and the role of open models in real AI systems.
1.7%at the 2025 report
8%a year earlier
Selected benchmarks, one year apart. The lead changes hands often and the gap has not closed in a straight line since.Stanford HAI, The 2025 AI Index Report (2025)
Key findings
Key figure
89%
89% of organizations that use AI use open source models somewhere in the stack, and 67% say open source AI is cheaper to deploy than proprietary alternatives. Smaller businesses adopt at higher rates than large enterprises.
Important context
Commissioned by Meta, which ships open-weight models. Read it as an interested party's survey.
Source: Linux Foundation Research, The Economic and Workforce Impacts of Open Source AI (2025)
Key figure
2.96M
Public model repositories on the Hugging Face Hub grew from 2.43M in January 2026 to 2.96M by August. Qwen alone accounts for 151,448 derivative models, 2.6× Meta's total footprint on the Hub.
Source: Hugging Face, State of Open Models: Summer 2026 (2026)
Topic 5 of 5: Environment
Emissions and water use
Environmental figures from company reports and independent estimates, with limitations noted.
GPT-4est.
5,184
Llama 3.1 405Best.
8,930
Grok 4est.
72,816
Grok 4, other estimateest.
~140,000
Outside estimates, not company disclosures. The 72,816 t figure is about what 17,000 cars emit in a year, and the AI Index carries a second Grok 4 estimate near double it. The spread is the honest state of the evidence.Stanford HAI, Inside the AI Index: 12 Takeaways from the 2026 Report (2026)
Key findings
Key figure
+81%
Google's 2025 emissions came in 18% above 2024 and 81% above its 2019 baseline, with electricity use up 37% year over year. Carbon-free energy stayed roughly flat at about 65%.
Source: Google, 2026 Environmental Report (2026)
Key figure
78%
Google replenished about 7.7 billion gallons of water, roughly 78% of the freshwater it consumed, short of the 120% replenishment goal it has set for 2030.
Source: Google, 2026 Environmental Report (2026)
Key figure
1.2M people
Inference water use for GPT-4o may exceed the annual drinking water needs of 1.2 million people.
Important context
Modeled from public figures; the operator has not published its own inference water total.
Source: Stanford HAI, Inside the AI Index: 12 Takeaways from the 2026 Report (2026)
Read the original reports
These screenshots were captured August 25, 2026. Open any report to compare our summary with the original publication.
Citing a problem is not the same as solving it. A small model on your own laptop does not fix the grid, and none of these reports were written about us. We collect them because they are the public record we are building against, and because you should be able to challenge every number here.
The version you can run yourself.
A small model, on your own machine, at its real speed. The demo answers from a box we host; the drive is $59.