Company hub
Microsoft
Every release, deal and policy move from Microsoft — updated daily.
TopicsStudies
-
Copilot Users Collaborate More, Not Less: Study of 12,000 Employees
Employees who adopted Microsoft 365 Copilot initiated about 1.4 more collaboration hours a month and reached about 2.2 more colleagues than never-adopters, a log study of 12,040 workers finds.
By
Model chronicle · Microsoft
MAI-Thinking-1 (7-model MAI family)
Flagship reasoning model plus MAI-Code-1-Flash, MAI-Image-2.5, MAI-Transcribe-1.5 and MAI-Voice-2.
microsoft.ai (source, opens in a new tab)MAI-1-preview & MAI-Voice-1
microsoft.ai (source, opens in a new tab)Phi-4
techcommunity.microsoft.com (source, opens in a new tab)Phi-3
azure.microsoft.com (source, opens in a new tab)
Sources: company announcements & release notes.
Fact box
- HQ
- Redmond, USA
- Founded
- 1975
- CEO
- Satya Nadella
- Core products
- Copilot · Azure AI Foundry · Phi · MAI models
- Stories
- 108
Sources: news.microsoft.com
Latest on X
Microsoft AI · @MicrosoftAI
Another No. 1 spot! MAI-Transcribe-2 takes first place for transcription accuracy in David AI's new benchmark. MAI-Transcribe-2 has the lowest error rate in 18 of 21 languages.
MAI-Image-2.6 & MAI-Image-2.6-Flash now available in Vercel!
MAI-Code-1.1-Flash is going local with on-device model calls inside Github Copilot. High-quality agentic coding on your device, with no inference charge for local model calls.
Build expressive voice agents with MAI models now in LiveKit. Learn more here: msft.it/6014alofr
MAI voice models are also available through OpenRouter. Learn more here: msft.it/6012aloIC
Editor's picks
Deals & funding
Microsoft · rounds, valuations, partners, acquisitions · latest 8 of 16
Partnership
Microsoft commits multiple billions to Mistral’s European GPU capacity
reuters.com (source, opens in a new tab)Partnership
Microsoft and OpenAI amend deal, ending exclusive licence
openai.com (source, opens in a new tab)Partnership
Microsoft and NVIDIA to invest up to $15B in Anthropic
Microsoft (up to $5B) and NVIDIA (up to $10B) to invest in Anthropic; Anthropic commits to $30B of Azure compute.
anthropic.com (source, opens in a new tab)Partnership
Lambda signs multibillion-dollar deal to deploy NVIDIA GB300 systems for Microsoft
lambda.ai (source, opens in a new tab)Partnership
IREN signs $9.7B five-year AI cloud contract with Microsoft
iren.com (source, opens in a new tab)Partnership
Microsoft and OpenAI sign new definitive agreement after recapitalisation
OpenAI recapitalises as a public benefit corporation; Microsoft holds about 27% (valued ~$135B) and OpenAI commits to $250B of Azure services.
openai.com (source, opens in a new tab)Partnership
Nscale contracts ~200,000 NVIDIA GB300 GPUs for Microsoft
nscale.com (source, opens in a new tab)Partnership
Microsoft adds Claude to Microsoft 365 Copilot
microsoft.com (source, opens in a new tab)
All 16 rounds & deals as a table · 0 funding, 16 partnerships, 0 acquisitions
| Date | Event | Type | Counterparty / round | Amount | Valuation | Source |
|---|---|---|---|---|---|---|
| Jul 21, 2026 | Microsoft commits multiple billions to Mistral’s European GPU capacity Microsoft will use Mistral Compute capacity in Europe and add Mistral models to Azure, Foundry and Copilot Studio; not a new equity investment (Reuters). | Partnership · compute | with Microsoft | — | — | reuters.com |
| Apr 27, 2026 | Microsoft and OpenAI amend deal, ending exclusive licence Amended terms: Microsoft stays primary cloud partner but loses exclusivity; OpenAI may sell products on other clouds. | Partnership · cloud | with OpenAI | — | — | openai.com |
| Nov 18, 2025 | Microsoft and NVIDIA to invest up to $15B in Anthropic Microsoft (up to $5B) and NVIDIA (up to $10B) to invest in Anthropic; Anthropic commits to $30B of Azure compute. | Partnership · compute | with NVIDIA, Anthropic | $15B | — | anthropic.com |
| Nov 3, 2025 | Lambda signs multibillion-dollar deal to deploy NVIDIA GB300 systems for Microsoft Multi-year agreement for tens of thousands of NVIDIA GPUs incl. GB300 NVL72; value not disclosed beyond 'multibillion'. | Partnership · compute | with Lambda | — | — | lambda.ai |
| Nov 3, 2025 | IREN signs $9.7B five-year AI cloud contract with Microsoft Access to NVIDIA GB300 capacity at IREN's Childress, Texas campus. | Partnership · compute | with IREN | $9.7B | — | iren.com |
| Oct 28, 2025 | Microsoft and OpenAI sign new definitive agreement after recapitalisation OpenAI recapitalises as a public benefit corporation; Microsoft holds about 27% (valued ~$135B) and OpenAI commits to $250B of Azure services. | Partnership · licensing | with OpenAI | $250B | — | openai.com |
| Oct 15, 2025 | Nscale contracts ~200,000 NVIDIA GB300 GPUs for Microsoft Deployments in Texas, Portugal, the UK and Norway; Reuters reported potential revenue of up to $14B. | Partnership · compute | with Nscale | — | — | nscale.com |
| Sep 24, 2025 | Microsoft adds Claude to Microsoft 365 Copilot Claude Opus 4.1 and Sonnet 4 become selectable in the Researcher agent and Copilot Studio alongside OpenAI models. | Partnership · integration | with Microsoft | — | — | microsoft.com |
| Sep 8, 2025 | Nebius signs $17.4B GPU infrastructure deal with Microsoft Five-year deal for capacity from Nebius' Vineland, New Jersey data center; can rise to $19.4B (Nebius filing). | Partnership · compute | with Nebius | $17.4B | — | reuters.com |
| May 19, 2025 | Microsoft hosts Grok 3 and Grok 3 mini on Azure AI Foundry Announced at Build: Grok models hosted and billed directly by Microsoft with Azure SLAs. Pricing from June 2025: Grok 3 $3/$15 and Grok 3 mini $0.25/$1.27 per 1M tokens (global). | Partnership · distribution | with Microsoft | — | — | devblogs.microsoft.com |
| Sep 20, 2024 | Constellation restarts Three Mile Island unit to power Microsoft (20-year PPA) 20-year power purchase agreement to restart the 835 MW Three Mile Island Unit 1 (renamed Crane Clean Energy Center) for Microsoft data centers. | Partnership · energy | with Constellation Energy | — | — | constellationenergy.com |
| Apr 16, 2024 | Microsoft invests $1.5B in Abu Dhabi’s G42 Minority stake; G42 runs its AI applications on Azure and Brad Smith joins G42's board. | Partnership · investment | with G42 | $1.5B | — | news.microsoft.com |
| Mar 19, 2024 | Microsoft hires Inflection AI founders, pays ~$650M licence Microsoft hires Mustafa Suleyman and most Inflection staff to form Microsoft AI; pays ~$650M mainly for licensing. | Partnership · licensing | with Inflection AI | $650M | — | blogs.microsoft.com |
| Feb 26, 2024 | Microsoft partnership: Mistral Large on Azure, €15M investment Multi-year deal: Mistral models on Azure AI, supercomputing for training. Microsoft's €15M (~$16M) minority investment converts into equity in a later round (reported by Reuters/FT). | Partnership · investment | with Microsoft | $16M | — | azure.microsoft.com |
| Jan 23, 2023 | Microsoft extends OpenAI partnership with reported $10B investment Third phase of the partnership: multiyear, multibillion-dollar investment; Azure remains OpenAI's exclusive cloud. | Partnership · cloud | with OpenAI | $10B | — | blogs.microsoft.com |
| Jul 22, 2019 | Microsoft invests $1B in OpenAI; Azure becomes exclusive cloud Microsoft invests $1B and becomes OpenAI's exclusive cloud provider; the two co-develop Azure AI supercomputing. | Partnership · investment | with Microsoft | $1B | — | openai.com |
Sources: company announcements & release notes, press reports.
-
Hybrid Computer-Use Agents Top Out at 41% When GUI and CLI Must Work Together
WeaveBench’s 114 real-world tasks: best PassRate 41.2%. GUI-only or CLI-only collapse below 3.5%. Outcome-only grading overestimates GPT-5.5 from…
-
AI Agents Guess Wrong at Critical Forks 40% of the Time, Taste-Bench Shows
Taste-Bench’s 502 decision-fork questions: the best frontier model scores only 59.7%. Later evidence is harder, and bigger reasoning…
More stories
Models & specs
Microsoft · context window, max output, API price, modalities, knowledge cutoff, license
| Model | Released | Context | Max output | Price in / out (per 1M) | Modalities | Knowledge cutoff | License / access | Spec source |
|---|---|---|---|---|---|---|---|---|
| MAI-Thinking-1 (7-model MAI family) | Jun 2, 2026 | — | — | — | — | — | — | microsoft.ai |
| MAI-1-preview & MAI-Voice-1 | Aug 28, 2025 | — | — | — | — | — | — | microsoft.ai |
| Phi-4 | Dec 12, 2024 | 16K | — | open weights | Text in → text out | — | Open weights, MIT 14B dense | techcommunity.microsoft.com |
| Phi-3 | Apr 23, 2024 | 128K | — | open weights 4K and 128K context variants. | Text in → text out | — | Open weights, MIT Phi-3-mini 3.8B (also small 7B, medium 14B) | azure.microsoft.com |
Sources: company announcements & release notes. Prices = standard API list price in USD per 1M tokens at time of release or current docs; batch is typically 50% off.