THE APEX TIMES
Microsoft to unveil Maia 300 AI chip in September as it presses TSMC for a larger order, reports Information
The next step in Microsoft’s custom AI silicon program is set for a September reveal, with the company reportedly negotiating with Taiwan Semiconductor Manufacturing Co. for capacity at a scale that would far exceed earlier Maia 200 production.
Microsoft is planning to reveal its next-generation AI chip, Maia 300, in September, according to a report carried by Yahoo Finance that cites The Information. The report frames the announcement as part of Microsoft’s broader effort to control more of the technology stack for artificial intelligence, including chips used to run AI workloads.
The same report says Microsoft is in talks with Taiwan Semiconductor Manufacturing Co. (TSMC) to secure supply for Maia 300. While Microsoft has already produced and deployed earlier Maia chips, the reporting indicates the newer Maia 300 order would be much larger than what has been produced so far for Maia 200.
In particular, the report characterizes Maia 200 production to date as being in the tens of thousands of chips, and it says the Maia 300 order being negotiated would dwarf that level. That comparison suggests Microsoft is preparing for a step-change in chip demand, likely tied to expected growth in AI training and inference workloads that require specialized accelerators.
For TSMC, the potential significance is straightforward: if Microsoft places a substantially bigger order for Maia 300 than it did for Maia 200, it would translate into more utilization tied to a high-value customer and a sustained custom-chip program. TSMC is already one of the world’s most important foundries for advanced semiconductor manufacturing, and customers that can reliably translate AI demand into multi-quarter volume can be particularly attractive.
Microsoft’s AI hardware roadmap has been shaped by the practical reality that cloud-scale AI pushes companies to seek better performance per watt, supply reliability, and cost control than commodity chips alone can offer. Custom silicon is also one way to reduce dependence on a single vendor’s general-purpose roadmap and to tailor compute to specific software stacks and model workloads.
Even so, the report does not provide detailed technical specifications for Maia 300, including the chip’s architecture, process node, power envelope, or performance targets. It also does not disclose expected production timing beyond the September reveal or the final size and commercial terms of the reported order with TSMC.
As with many chip supply stories, the biggest uncertainties remain the exact manufacturing schedule and the firm’s final commitment level. The report centers on negotiations and a planned reveal, but it does not lay out whether Maia 300 will ramp immediately after the announcement, how much capacity will be reserved, or how the chip will be allocated across Microsoft’s internal cloud and customer workloads.
Why It Matters
- A Maia 300 ramp would announcement whether Microsoft’s custom silicon strategy is moving from early deployments to a larger, volume-driven phase.
- Bigger reported ordering talks with TSMC suggest Microsoft may be securing capacity to meet projected AI demand and avoid bottlenecks.
- If custom chips materially expand, the balance between specialized accelerators and general-purpose GPUs could shift across cloud AI workloads.
- Chip reveal timing and supply commitments can influence expectations about when Microsoft can scale AI services at lower cost or higher efficiency, though specifics were not disclosed in the report.
Key Facts
- Microsoft plans to reveal its Maia 300 AI chip in September, according to a report cited by Yahoo Finance that references The Information.
- The report says Microsoft is negotiating with TSMC for supply for Maia 300.
- The report characterizes Maia 200 chips produced to date as being in the tens of thousands.
- The report says the Maia 300 order being discussed would be much larger than Maia 200 production levels.
- Microsoft’s custom AI chip program is positioned around scaling AI compute through specialized hardware.
Technology Related
Amazon receives a warrant tied to $521.4 million of 3M stock as data-center partnership expands
The expanded arrangement links Amazon’s future purchases of 3M shares to the hyperscaler’s data center demand, Reuters-style market reporting said as MMM moved higher after the announcement.
Citi Pins the “AI Debt Squeeze” Selloff on Rising Bond Yields, Not Demand, and Backs Microsoft
In a note cited by Yahoo Finance, Citi analyst Tyler Radke argues that the pressure on “neocloud” companies is being driven more by borrowing costs than by weakening appetite for AI-related infrastructure, and he favors Microsoft over smaller cloud peers.
3M shares rise after report ties Amazon data-center spending to potential stake in the industrial company
A market report says Amazon could gain an ownership position in 3M if it spends $6 billion on 3M’s data-center products, as the industrial supplier leans further into AI-related infrastructure and a planned Taiwan investment.
JPMorgan upgrades Intel to Neutral in fresh Wall Street note, according to Yahoo Finance
A market wrap flagged JPMorgan’s constructive stance on Intel shares, lifting its rating to Neutral. The brief update offered limited detail on what drove the shift.
Marathon Petroleum’s surge puts a dent in Nvidia’s market lead as one analyst flags more upside
A sharp run in Marathon Petroleum shares has pushed the U.S. refiner above major semiconductor names in 2026 performance, drawing attention to how quickly leadership can shift across sectors.
Netflix highlights “anime classics” in a new newsroom push aimed at keeping older fandoms engaged
In a post on its Netflix Newsroom, the streaming company spotlights well-known anime titles as a way to draw in viewers who may not yet have explored the genre’s earlier waves.
Broadcom signs a 10-year VMware software deal to expand NTT DOCOMO’s private cloud in Japan
Broadcom says it will help NTT DOCOMO move forward with a private cloud built on VMware Cloud Foundation under what the company describes as its first 10-year VMware software agreement in Japan.
Netflix joins GM and United Airlines on Tuesday’s earnings watchlist, as investors also track SpaceX-related headlines
A market roundup ahead of the Oct. 20 session highlighted earnings from Netflix, General Motors and United Airlines, with SpaceX among the other items set to draw attention from traders.
Oracle drops about 9% as tech sells off and investors weigh AI infrastructure financing
The pullback in Oracle shares follows a broader market retreat and comes after hotter September inflation figures, while traders also focused on the costs and funding of the company’s AI buildout.
Nvidia’s AI surge meets a wary counterpoint: customers may seek less dependence on its chips
A fresh market commentary argues Nvidia’s valuation upside could run into a risk if major buyers start designing roadmaps that reduce how much they rely on a single supplier. The point is less about demand collapsing and more about bargaining power and architectural switching.