Microsoft has introduced MAI Code 1.1 Flash, a cheaper and faster successor to its proprietary coding model, and deployed it on GitHub Copilot as it pushes to lower the cost of AI development tools and the infrastructure behind them. The model, announced on the 11th local time, is priced at $0.20 per million input tokens, $1.20 per million output tokens, and $0.02 per million cached input tokens, marking roughly a 73% reduction from the prior version after Microsoft refined training and service operations. Internal tests showed Terminal Bench 2.1 gains of 22%, .NET task improvements of 15%, token generation that is 25% faster, and a 25% drop in the number of tokens needed to complete tasks. In GitHub Copilot usage, Microsoft said generated-code retention rose 4% and user return rates increased 9%. The model also adds image understanding and is available for paid users in Visual Studio Code and Copilot CLI (command-line interface), while free and student users can access it when Copilot automatically selects the best model. Support for MAI Code 1 Flash will end on September 10th. The software push came alongside signs that Microsoft is moving ahead with Maia 200, a custom inference chip (processor optimized for running AI models) that Barron's said could arrive as early as September. Microsoft has not officially confirmed that timeline or production scale, but the chip strategy points to a broader effort to control AI infrastructure costs. Maia 200 uses TSMC's 3-nanometer process and is described as having more than 140 billion transistors, 216 GB of high-bandwidth memory (HBM), and more than 10 petaflops at 4-bit computation, with 30% better performance per dollar than the latest alternative hardware Microsoft previously used. Deployment has begun in Iowa, with Arizona and other regions planned next. The stakes are high because Copilot, Azure, and OpenAI workloads require vast computing power, and Microsoft expects roughly $50 billion in capital expenditures this quarter. Its shares closed down about 0.6% at $502.61 on the 11th, and that price was about 12.69% below the GF Value estimate of $575.66 cited by some analysts. The investment case now turns less on chip specifications than on whether Maia can reduce the cost of every AI token enough to improve margins and justify the spending.