Models & Labs
Microsoft Aims to Develop In-House AI Models
Matt Wolfehigh confidence
Why it matters
- →Indicates a strategic shift in Microsoft's AI development approach.
- →Suggests potential competition with OpenAI in AI model creation.
- →Could lead to new, proprietary AI technologies from Microsoft.
At Microsoft Build 2026, Microsoft AI CEO Mustafa Suleyman announced the company's intention to develop AI models entirely in-house. This marks a strategic shift from their previous approach of integrating OpenAI's models into Microsoft products. Suleyman emphasized the goal of creating the 'absolute frontier' of AI models, highlighting a new direction for Microsoft's AI strategy.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1OpenAI Unveils New Spud AI Model
AI Explained · March 26, 2026 · Background
- 2Microsoft to unveil new AI models at Build
The Verge AI · June 1, 2026 · Related
- 3Microsoft Introduces Scout, an AI Office Assistant
WIRED AI · June 2, 2026 · Background
- 4Microsoft Unveils MAI-Thinking-1 AI Model
The Verge AI · June 2, 2026 · Related
- 5Microsoft Unveils AI Models and Agentic Hardware at Build 2026
The Rundown AI · June 3, 2026 · Related
- 6Microsoft's MAI Models Target Enterprise AI Efficiency
The AI Daily Brief · June 4, 2026 · Related
- 7Microsoft Launches Seven New MAI Models
Matt Wolfe · June 5, 2026 · Related
- 8Microsoft Unveils 7 New AI Models at Build
Matt Wolfe · June 8, 2026 · Same story
What happened next
- 9Microsoft sells OpenAI models in China
AI News · June 18, 2026 · Background
- 10Microsoft Shifts to In-House AI Models for Cost Savings
TechCrunch AI · July 7, 2026 · Same story
- 11Microsoft Trains Sales Team to Critique AI Rivals
TechCrunch AI · July 15, 2026 · Related
- 12Microsoft Develops Cheaper In-House AI Models
The AI Daily Brief · July 25, 2026 · Same story
- 13OpenAI's Full-Stack Approach to AI Development
OpenAI · July 31, 2026 · Background
Follow this story
Open the full story →Microsoft pivots to independent AI development
2 developments
- Jun 3 · The Verge AI
Microsoft pivots to independent AI development - Jun 9 · Matt Wolfe
Microsoft Aims to Develop In-House AI Models (This article)↳ Microsoft plans to build AI models from scratch, moving away from reliance on OpenAI.
More in Models & Labs
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
llama.cpp Releases
Models & Labscoding
llama.cpp b12342 adds few-row MMQ for speculative decoding
Speculative decoding just got a significant speed boost on NVIDIA GPUs. This release introduces a specialized kernel for 'few-row' matrix multiplications, targeting the small batch sizes typical of verification passes in speculative decoding workflows. By optimizing these specific operations for quantized formats like Q8_0 and IQ4 variants, llama.cpp reduces latency where it matters most during token generation. The change is CUDA-only for now but leverages hardware capabilities from Ampere onwards, making local inference faster without requiring new model architectures or complex configuration changes.
llama.cpp Releases
© TechCrunch AIModels & Labsmodels
Google deploys Gemini 4 Hydrogen for faster Search
Google is quietly shifting its search infrastructure by replacing older fast models with Gemini 4 Hydrogen, a tier designed specifically for speed and cost-efficiency rather than raw reasoning power. This model completes tasks 2.5 times faster than its predecessor while beating the more powerful Argon on specific intelligence benchmarks, proving that specialized efficiency can outperform brute force in high-volume workflows. By integrating this into Search's fan-out technique, Google now breaks down complex queries into subqueries with greater precision and speed, delivering richer visual responses without the latency of heavier models. This marks a strategic pivot toward agentic execution layers within consumer search, making AI Mode feel snappier for everyday multi-part questions.
TechCrunch AI