Company tracker 6 of 10 · 10 companies, one lens
Microsoft
Microsoft, maker of Copilot and the Phi models, tracked through the same six lenses as every company on AI Adoption.
Its best model, Phi-4, scores 130.5 on the Epoch Capabilities Index, #10 of 10, 38.3 behind Anthropic. In March 2026, 10.5% of US adults said they had used Microsoft Copilot in the past week. It released 0 notable models in 2025, #10 of 10 by count. It owns AI chips equal to about 3.32 million H100s.
4 ledger entries0 charts name MicrosoftLatest entry Jul 2026Checked 20 Oct 2026
Microsoft through six lenses
How the lenses work →The same six lenses, in the same order, on every company page. Each shows one comparable number where a source measures several companies the same way, its rank, and when it last changed.
01Capability#10 of 10
How good are its best models?
130.5
Epoch Capabilities Index score of its best model, Phi-4 (Dec 2024)
Epoch combines many benchmarks into one scale. 90% interval for this model: 124.7 to 132.0.
02Output#10 of 10
How much does it ship?
0
Notable AI models released in 2025
42 notable models from 2014 to 2025 and 2 in 2026 so far. Model record →
03Money
How much does it earn?
Not measured yet
No AI revenue figure for Microsoft in Epoch AI's company data
Epoch AI tracks revenue for AI-first companies such as OpenAI, Anthropic, Mistral AI, DeepSeek.
04Compute#9 of 10
How much compute and energy goes into its AI?
8.6×10²³
FLOP: its largest training run with a published estimate, Megatron-Turing NLG 530B (Oct 2021)
Largest estimated training cost: $4M (Megatron-Turing NLG 530B), Epoch AI. Model record →
Ranked among the 10 tracked companies. Only runs with a published estimate count, so a company that discloses less can rank lower.
Owns AI chips equal to about 3.32M H100s (Mar 2026), #2 of 7 named owners. Epoch AI data →
05Use#3 of 7
Who uses it, and for what?
10.5%
of US adults said they used Microsoft Copilot in the past week, March 2026
Epoch AI with Ipsos, a probability-based sample of 2,021 US adults; 49.2% used no AI service. Among employed adults in July 2026: 14.8%.
06Safety & policy
How reliable is it, and how is it governed?
Not measured yet
Not on the reliability test
- Published safety framework, and each version
- Third-party pre-release evaluations
- Regulatory actions and fines
The ledger
What gets an entry →An entry goes in only if it changes a number or a position on one of the six lenses. If it moves nothing, it is not news here. Newest first.
July 2026
- Jul 2026UseNew14.8% of employed US adults said they had used Microsoft Copilot in the past weekProbability-based sample of 1,103 employed US adults, asked about work and personal use.
March 2026
- Mar 2026ComputeNewOwned AI chips equal to about 3.32 million Nvidia H100sEpoch's 90% range: 2.56M to 4.30M. #2 of the 7 owners Epoch lists by name.3.32M
- Mar 2026UseNew10.5% of US adults said they had used Microsoft Copilot in the past week#3 of the 7 services asked about that belong to tracked companies. 49.2% had used no AI service. Probability-based sample of 2,021 US adults.#3
December 2024
- 12 Dec 2024CapabilityNewPhi-4 scored 130.5 on the Epoch Capabilities Index, its best model#10 of the 10 tracked companies' best models; Anthropic's best scored 168.8. Epoch's 90% interval: 124.7 to 132.0.#10
No entries match. Try another lens.
Microsoft's model record
Download the CSV →Every notable model Epoch AI has logged for Microsoft: 47 in all, 42 from 2014 to 2025 and 2 in 2026 so far. Training compute, cost and hardware are Epoch's estimates, shown only where it has one.
Notable models, 2014–202542#4 of 10 companies · Epoch AI's count, as in chart 1.1.7
Frontier models3#5 of 10 companies · marked frontier in Epoch AI's database
Open weights33%2 of 6 notable models since 2023
Largest estimated training cost$4M#9 of 10 companies · Megatron-Turing NLG 530B (Oct 2021), 2023 dollars
Largest training cluster4,480#6 of 8 companies · NVIDIA A100 SXM4 80 GB · Megatron-Turing NLG 530B
Notable models per year* 2026 to 19 Oct 2026
Notable models since 2023, newest first6 models
ReleasedModelAccessTraining computeEst. costTraining hardware
2 Jun 2026MAI-Image-2.5API or hosted access———
2 Jun 2026MAI-Code-1-FlashAPI or hosted access———
5 Nov 2023LLaVA 1.5Open weights7.8×10²²—8 NVIDIA A100
26 Oct 2023CODEFUSION (Python)Unreleased7.9×10¹⁸$0K4 NVIDIA V100
17 Apr 2023LLaVAOpen weights7.8×10²²$0K8 NVIDIA A100
5 Jan 2023VALL-EUnreleased1×10¹⁹$0KNVIDIA V100
Data: Epoch AI, Data on AI Models, snapshot 19 Oct 2026, CC BY 4.0. Organisations are grouped into companies as in chart 1.1.7 (for example, Google DeepMind and Google Brain count as Google). Microsoft's models as CSV.
Microsoft beside the others
All 10 companies →The same rows for every company. A dash means no source measures it for that company yet: gaps are shown, never filled with a guess. The small number is the rank among the companies in the full table that the source covers.
CapabilityEpoch Capabilities Index, its best model · data130.5#10168.8#1168.8#1166.4#3156.6#6161.8#4155.9#7157.4#5150.0#8146.0#9
MoneyAnnualized revenue, latest report · data—$50B#2Sep 2026$65B#1Jul 2026—$428M#5Sep 2025—$1B#3Sep 2026—$1B#3Sep 2026—
ComputeLargest training run with a published estimate, FLOP · chart8.6×10²³#9Megatron-Turing NLG 530B, Oct 2021 (with Nvidia)9.4×10²⁶#1GPT-6 Astra, Sep 20263.4×10²⁵#5Claude 3.7 Sonnet, Feb 20255×10²⁵#4Gemini 1.0 Ultra, Dec 20235×10²⁶#2Grok 4, Jul 20255.2×10²⁵#3Llama 4 Behemoth (preview), Apr 20259.7×10²⁴#8DeepSeek-V4-Pro, Apr 20261.5×10²⁵#7Qwen3-Max, Sep 20257.7×10²³#10Mixtral 8x7B, Dec 20231.8×10²⁵#6Nemotron-4 340B, Jun 2024
UseUS adults who used its assistant in the past week, Mar 2026 · data10.5%#331.0%#13.0%#620.9%#24.5%#57.5%#40.6%#7———
Safety & policyLowest hallucination rate of its models (lower is better) · chart—89%#826%#250%#322%#187%#782%#489%#884%#683%#5
More measures · Epoch AI data
Model record · Epoch AI estimates
Notable models, 2014–2025Epoch AI's count, as in chart 1.1.742#461#313#7194#17#887#27#827#66#1030#5
Frontier modelsmarked frontier in Epoch AI's database3#515#23#533#13#55#30#80#80#84#4
Open weightsshare of its notable models since 202333%2 of 64%2 of 570%0 of 2214%8 of 5711%1 of 973%19 of 26100%10 of 1075%24 of 3267%4 of 683%15 of 18
Largest estimated training cost2023 US dollars, among models Epoch has costed$4M#9Megatron-Turing NLG 530B$366M#2GPT-4.5$26M#5Claude 3.5 Sonnet$31M#4Gemini 1.0 Ultra$388M#1Grok 4$53M#3Llama 3.1-405B$7M#8DeepSeek-R1 (May 2025)$109K#10PLUG$14M#7Mistral Large$21M#6Nemotron-4 340B
Largest training clusterchips for one model, where Epoch records it4,480#6Megatron-Turing NLG 530B100,000#2GPT-6 Astra—57,000#3Gemini 1.0 Ultra200,000#1Grok 432,000#4Llama 4 Behemoth (preview)2,048#7DeepSeek-V3 (Mar 2025)512#8CogView—6,144#5Nemotron-4 340B
Charts on AI Adoptionthat name the company0332223844714
Every chart that names Microsoft
All charts →0 charts across 0 hubs name Microsoft or one of its models. Each has its source and a CSV of its data.
Follow Microsoft
The AI numbers worth knowing, every Friday.
The week in AI numbers is one email on Fridays with the new charts across every theme, each linked to its source. Following Microsoft also tells us which companies you want to hear more about.
You're subscribed. Manage your email
Free, one email a week. Choose what you follow →The other companies, same lens
AI Companies →OpenAI168.8Epoch Capabilities Index65 ledger entriesOpen tracker →Anthropic168.8Epoch Capabilities Index61 ledger entriesOpen tracker →Google166.4Epoch Capabilities Index48 ledger entriesOpen tracker →xAI156.6Epoch Capabilities Index37 ledger entriesOpen tracker →Meta161.8Epoch Capabilities Index33 ledger entriesOpen tracker →DeepSeek155.9Epoch Capabilities Index35 ledger entriesOpen tracker →Alibaba157.4Epoch Capabilities Index33 ledger entriesOpen tracker →Mistral AI150.0Epoch Capabilities Index26 ledger entriesOpen tracker →Nvidia146.0Epoch Capabilities Index7 ledger entriesOpen tracker →
How the ledger works
- One lens for everyone. Six lenses, same order, same definitions, for every company. No company gets an extra lens for a product it wants to talk about.
- Changes, not news. An entry must move a number or a rank on a lens. Launch events, interviews and hires do not count unless they do.
- Evidence tier on every line. Independent measure, independent estimate or company data. A company's own claims are replaced when an independent score arrives.
- Gaps stay visible. If no source measures a lens for a company, the page says so.
Disclosure: AI Adoption uses AI tools in its research and editorial work, including products from some of the companies tracked here. No company has any say in its page or its ledger, and the same rules apply to every company. Editorial standards · Data sources