Story perspectives

Microsoft launches 35B AI model, faster Web IQ, and quantum chip

6/3/2026

Summary

Microsoft Unveils AI

Microsoft unveiled MAI-Thinking-1, a 35-billion-parameter model with a 128,000-token context window that scored 97% on AIME and offers up to 10× better cost efficiency.
Microsoft presented Majorana 2, a quantum chip that boosts qubit reliability 1,000-fold, reaches a mean lifetime of 20 seconds, and targets commercial deployment by 2029.
Microsoft unveiled Web IQ, a grounding API that runs about 2.5 times faster than rivals and powers Microsoft Copilot and OpenAI’s ChatGPT.

Summary

Microsoft Launches Badge

Microsoft’s Applied Sciences Group built the Solara prototype in three days with Qualcomm and MediaTek chips.
The Android badge includes a touchscreen, fingerprint sensor, privacy switch, side-camera, Wi-Fi, Bluetooth and 5G and will be licensed to partners instead of sold.
Critics warn that continuous cloud streaming of audio and video raises privacy risks, while Nadella activated the badge’s fingerprint sensor, captured audience photos and sent them instantly for AI agents.

Summary

Microsoft Launches DevBox

Microsoft launches the Surface RTX Spark Dev Box to cut developers’ cloud spend and reshape Azure revenue.
The Dev Box, pre-installed with a developer stack and in a 3-D-printed chassis, bundles Azure management services to lock enterprises into a seamless on-premise-and-cloud workflow.
Challenging Apple’s Mac Mini in the desktop segment, the offering prompts analysts to warn that the “unmetered intelligence” tier could shift AI pricing from per-token to fixed-cost hardware.

Summary

Microsoft Bans Claude

Microsoft told employees in certain divisions to stop using Anthropic’s Claude Code and switch to its own Copilot.
The mandate collects employee interaction data to tune models for VS Code and GitHub, while staff view it as a cost-cutting removal of a preferred tool.
Suleyman said Microsoft will avoid shortcuts like model distillation, opting for custom chips and in-house AI to pursue its lowest-cost AI provider goal.

History

Historical Context

Nvidia’s move toward programmable graphics began with the 1999 GeForce 256, the first GPU marketed with on-chip transformation and lighting, and a $200 million Xbox hardware contract in 2000 that focused its engineers on compute. Decades later Microsoft’s Applied Sciences Group follows that path, deploying custom Majorana 2 silicon on Azure to bind developers to its cloud AI services.

1 of 5

Microsoft Unveils AI

Microsoft unveiled MAI-Thinking-1, a 35-billion-parameter model with a 128,000-token context window that scored 97% on AIME and offers up to 10× better cost efficiency.
Microsoft presented Majorana 2, a quantum chip that boosts qubit reliability 1,000-fold, reaches a mean lifetime of 20 seconds, and targets commercial deployment by 2029.
Microsoft unveiled Web IQ, a grounding API that runs about 2.5 times faster than rivals and powers Microsoft Copilot and OpenAI’s ChatGPT.

1 / 5