Google has joined Amazon in attacking Nvidia’s stronghold with AI app developers
THE SO WHAT
Google and Amazon both shipping tools to move PyTorch workloads off Nvidia GPUs is a clear attempt to loosen CUDA lock-in at the developer level. If you’re building heavy AI workloads, treat portability as a design requirement now—abstract your infra so you can arbitrage accelerators as alternatives mature.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIPartnering with Blacksmith
If agents can write production-scale code, the new bottleneck is proving that code is safe, performant, and compliant at the same speed. Expect “agent-era CI/CD” vendors like Blacksmith to become a required layer in any serious agent deployment, not a nice-to-have devtool.
Applied AIAnthropic S1 Shows $80 Billion Loss This Year, WSJ Reports
An $80.7B loss and a projected $750B compute bill reframes frontier AI as mega-project infrastructure, not software economics. For enterprise buyers, this level of capex concentration means vendor risk and pricing power need to be part of your AI strategy, not an afterthought.
Applied AIAnthropic May Spend $750 Billion on Computing Costs, WSJ Reports
A projected $750B compute tab concentrates enormous power in whoever controls advanced fabs, energy, and AI cloud capacity. Operators should assume AI pricing, access, and even feature roadmaps will be constrained by upstream infra deals you don’t control—diversify vendors and keep optionality where you can.
Applied AIGoogle is bringing a more personal version of its Gemini AI assistant to your car
Gemini pulling from Gmail, Calendar, Maps, and YouTube in-car turns the vehicle into another deeply personalized assistant surface. Auto and mobility players now have to assume the OS owner intermediates driver relationships—design services and partnerships that survive being one layer down from the assistant.