48 stories on Microsoft since Apr 29, 2026, newest first.
Microsoft's third-gen Maia 300 chip uses TSMC N2P and 288GB HBM4 to challenge Nvidia's grip on AI inference, targeting multi-gigawatt Azure scale.
Microsoft's Maia 300 chip delivers 40% more tokens per megawatt than leading GPUs and 7x the throughput of Maia 200, targeting full production in 2027.
Microsoft's first Code of Conduct for MAI models bars AI from resisting shutdown, hiding reasoning, or widening its own scope without human direction.
Satya Nadella argues proprietary AI feedback loops build durable advantage, while enterprises that only rent frontier model access will fall behind.
Microsoft Agent 365 brings AI agent governance to all 276,000 KPMG staff, marking the Big Four's first full enterprise-wide AI agent deployment.
Microsoft triples its Japan pledge to $10B, partnering with Sakura Internet to build sovereign Azure AI compute for government and enterprise through 2029.
Microsoft's Majorana 2 chip claims 1,000x more reliable qubits with a 20-second lifetime, targeting a commercial quantum computer by 2029.
Microsoft MAI-Code-1-Flash, a 5B in-house coding model, beats Claude Haiku 4.5 on SWE-Bench Pro and uses up to 60% fewer tokens in GitHub Copilot.
Microsoft launched 7 MAI models spanning code, reasoning, image, voice, and transcription at Build 2026, a full-stack bid to cut its OpenAI reliance.
Anthropic is in talks to run Claude inference on Microsoft Maia 200 chips via Azure, a 30% cheaper bet that loosens its dependence on Nvidia GPUs.
Microsoft Intelligent Terminal launches as an open-source AI agent shell backing Claude Code, Codex, and Gemini, taking on Warp for developers in 2026.
Microsoft Agent Store pays developers an 85 percent revenue share, triple the Apple and Google cut, to win the AI agent race inside 400M Windows seats.
Microsoft Polaris replaces GPT-4 Turbo as the default GitHub Copilot model in August 2026, running on in-house Maia chips and cutting OpenAI reliance.
Microsoft unveiled Project Solara at Build 2026, an Android platform for enterprise devices that run AI agents instead of apps, piloted by CVS and Target.
Microsoft's Frontier Tuning lets enterprises train custom MAI models on proprietary data with RL, keeping the weights inside their own infrastructure.
Microsoft MXC enforces AI agent permissions at the Windows kernel, booting 94% faster than a Hyper-V VM, with OpenAI and NVIDIA on board.
Nvidia DGX Station for Windows runs 1 trillion-parameter models locally on the GB300 superchip, a deskside bet that frontier AI does not need cloud.
Microsoft MAI-Thinking-1 is its first reasoning model trained with zero OpenAI data, scoring 97% on AIME 2025 and matching Claude Opus 4.6 on SWE-Bench.
Azure Agent Mesh federates AI agents across on-prem, Cloud PCs, and edge under one governance layer, with consumption-based pricing and Q4 2026 GA.
Microsoft Foundry Local hits general availability on Windows, macOS, and Linux, running AI on-device with no cloud round trip and no per-token bill.
Microsoft Agent 365 adds a registry that finds unmanaged AI agents via Defender, Entra, and Intune, extending identity control to non-human workers.
Microsoft IQ unifies Work IQ, Fabric IQ and Web IQ at Build 2026, giving AI agents live M365 and business data context for autonomous action.
Microsoft Scout, the first Autopilot agent at Build 2026, monitors your inbox and Teams 24/7 using Work IQ context to act without being asked.
Microsoft's Surface Laptop Ultra delivers 1 petaflop of AI on Nvidia RTX Spark with 128GB RAM, targeting AI developers who build on Windows.
GitHub Copilot's new desktop app manages fleets of autonomous agents, with Enterprise customers getting hands-free coding starting July 2026.
Microsoft's MAI-Thinking-1 matches Claude Sonnet 4.6 in blind tests at 10x lower cost, signaling a new phase of AI model competition.
Azure AI Foundry's Agent Orchestrator preview launches in August 2026 to balance load across thousands of AI agents for enterprise deployments at scale.
Microsoft's Aion 1.0 Plan brings a 14B-parameter reasoning model into Windows 11, enabling local agentic workflows without cloud latency or API costs.
Microsoft open-sources Windows Agent Framework at Build 2026, giving developers a vendor-neutral platform for building AI agents on any model.
Microsoft wins a $9.69 billion DoD contract to deploy Copilot AI across all US military branches, the largest Microsoft government deal in history.
Microsoft flipped Office 365 Copilot to Agent Mode by default at Build 2026, governing agents across AWS and Google clouds via its new Agent 365 plane.
Microsoft removed the AGI clause that could end its OpenAI license, fixing the deal through 2032 and handing the AGI verdict to an independent panel.
Microsoft used Build 2026 to open-source its Windows Agent Framework and name Project Polaris to replace GPT-4 in GitHub Copilot by August.
Microsoft committed $10B to Japan through 2029 for AI data centers, sovereign cloud, and training 1 million engineers, its largest bet on the country yet.
Anthropic is in talks to run Claude on Microsoft's custom Maia 200 AI chips via Azure, seeking 30% cost efficiency gains over rival silicon.
GitHub Copilot's token-based billing went live June 1, replacing flat-rate plans with per-token costs that could double expenses for developers running agentic coding workflows.
Amazon, Google, Meta, and Microsoft will spend a combined $725 billion on AI infrastructure in 2026, up 77% from 2025, straining the global power grid.
Microsoft's Copilot Cowork, built on Anthropic's Claude, marks a multi-model pivot that threatens OpenAI's grip on 300 million enterprise users and redefines who owns the enterprise AI stack.
Korea's generative AI adoption hit 31.7% in Q1 2026, the fastest single-period growth of any country, driven by improved Korean-language models and national policy support.
Microsoft released MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2 through Azure Foundry, outperforming OpenAI's Whisper on all 25 languages and entering the voice cloning and image generation markets.
Microsoft data: AI adoption hit 17.8% of global workers in Q1 2026, but the developed gap widened by 1.5 points and 82% have never used an AI tool.
On April 27, 2026, OpenAI and Microsoft ended their exclusive partnership, capping revenue share and freeing OpenAI to sell on any cloud. The implications go far beyond the press release.
Microsoft Agent 365 goes GA May 1, 2026 at $15/user/month, a cross-cloud governance control plane for AI agents backed by Entra, Defender, and Purview.
Microsoft's Legal Agent for Word brings AI-powered contract review and redlining to legal professionals via Microsoft 365 Copilot Frontier preview.
Microsoft's open-source Agent Governance Toolkit covers all 10 OWASP Agentic AI risks at under 0.1ms latency, MIT licensed, released April 2, 2026 ahead of EU AI Act obligations.
Microsoft announced a $10 billion, four-year investment in Japan's AI infrastructure, pledging to train 1 million workers and partner with SoftBank and Sakura Internet to keep sensitive AI workloads in-country.
Amazon Quick launches as AWS brings GPT-5.5, Codex, and OpenAI Managed Agents to Bedrock via a $38B deal, ending Microsoft's seven-year exclusive lock on OpenAI models.
OpenAI ends its 7-year cloud exclusivity with Microsoft, freeing models to ship on AWS and Google Cloud as Amazon seals a $50B investment deal