Microsoft’s next-generation AI chip, code-named “Braga,” faces a significant setback with mass production delayed until at least 2026—a six-month minimum postponement from its original 2025 target. The delay, attributed to design changes, staffing limitations, and high employee turnover, deals a blow to Microsoft’s strategy to reduce dependence on Nvidia’s dominant AI hardware. Worse yet, internal projections indicate Braga will underperform Nvidia’s Blackwell chip when it finally launches. As reported by The Information, this development cedes ground to rivals like Google and Amazon, both advancing proprietary AI chips faster.
Microsoft’s Braga AI Chip Design and Production Challenges
Unanticipated design modifications—reportedly influenced by evolving requirements from partners like OpenAI—forced a retooling of Braga’s architecture. Compounding this, staffing shortages and unusually high turnover within Microsoft’s chip division hampered progress. Three sources directly involved confirm these issues created a “perfect storm” delaying production. Microsoft’s silence on the matter (evidenced by repeated non-responses to Reuters and other outlets) underscores the sensitivity of the setback.
Braga Performance Shortfalls Against Nvidia
When Braga reaches production, it is projected to “fall well short” of Nvidia’s Blackwell B200 in key benchmarks. Blackwell, launched in late 2024, already outperforms its predecessor (H100) by 33-57% in computer vision tasks and boasts 2.4× higher memory bandwidth. This performance gap is strategic: Braga aimed to cut Azure’s AI infrastructure costs and boost efficiency. Delaying its rollout forces Microsoft to keep paying premium prices for Nvidia GPUs, squeezing Azure margins.
Competitive Landscape Widens
While Microsoft stalls, rivals accelerate:
-
Google launched its 7th-generation Tensor Processing Unit (TPU) in April 2025, optimizing AI application speed.
-
Amazon unveiled Trainium3 in December 2024, set for late-2025 deployment.
Both leverage custom silicon to reduce costs and bypass Nvidia. Microsoft’s earlier Maia 100 chip (November 2023) already lagged, focusing on image processing instead of generative AI. Braga’s delay exacerbates this innovation gap.
Microsoft’s AI Stack
The delay undermines Microsoft’s end-to-end AI stack ambitions. Azure’s growth faces headwinds as competitors integrate cost-efficient custom chips into their clouds. Financially, Microsoft remains tethered to Nvidia’s pricing power—where self-hosting a Blackwell cluster costs $0.51/GPU/hour versus up to $16.10 for cloud-based alternatives. Analyst warnings highlight “risk to Azure’s profit margins” and a “multi-year lead” for Nvidia.
Investor and Market Reactions
Nvidia’s stock (up 35% YTD) contrasts with Microsoft’s flat performance, reflecting market confidence in AI hardware leaders. The Braga delay intensifies pressure on Microsoft to demonstrate AI infrastructure independence. With Nvidia’s next-gen GB300 chip due in 2026—promising 35× lower inference costs—Microsoft’s window to compete narrows.
Broader Industry Context
Big Tech’s race for custom AI chips stems from soaring Nvidia GPU costs and supply constraints. Microsoft, Amazon, and Google collectively invested billions to bypass these hurdles. Yet Microsoft’s execution struggles highlight the complexity of semiconductor development—where design agility and talent retention prove critical. As one report bluntly stated: “In AI, speed isn’t just a feature—it’s survival. Microsoft just got outrun.”
Similar Posts
- Microsoft Edge Beta 138 Introduces Advanced AI-Powered History Search and Media Control: Complete Guide and Privacy Insights
- Microsoft’s Windows Resiliency Initiative Brings Quick Machine Recovery, Connected Cache, and Secure Printing to Enterprises
- Windows 11 Insider Preview Build 26200.5641 (Dev Channel) Released: New and Improved Start Menu, Lock Screen Widgets, and More
- Convenient Microsoft Authenticator Autofill Feature Set for Early Retirement by August 2025—Users Urged to Migrate to Edge Instead
- Microsoft Store Gets Faster, More Personalized, and Injected with Copilot AI: A Deep Dive into the Latest Windows 11 Update
Discover more from Microsoft News Today
Subscribe to get the latest posts sent to your email.
