Microsoft orders 300,000+ Maia 300 AI chips from TSMC to cut Nvidia reliance

Microsoft Maia 300 Launch Signals Major Push Into Custom AI Silicon

Microsoft is preparing to unveil its next-generation Maia 300 AI chip this fall, potentially as early as September, marking a significant escalation in the company's efforts to reduce dependence on Nvidia1

1

2

2

. The tech giant has been negotiating with TSMC to secure manufacturing capacity for more than 300,000 Maia 300 chips targeted for delivery in 20271

1

3

3

. Andrew Wall, general manager for Microsoft's Azure Maia, confirmed that the company "continues to invest in custom silicon as part of our long-term AI infrastructure strategy," though he noted the reported production figures "don't reflect the scale of our program"2

2

.

The scale of Microsoft's ambition extends beyond the initial 300,000-unit order. A company manager stated the ultimate goal is to produce "gigawatts' worth" of Maia chips, with longer-term targets exceeding one million units1

1

3

3

. This represents a fundamental shift in Microsoft's AI infrastructure strategy, moving from purchasing compute capacity to building its own AI compute layer.

Racing to Close the Gap With Amazon and Google

Microsoft's custom AI silicon program faces a critical challenge: it started years after Amazon and Google began designing their own AI accelerators1

1

. Google has already started booking revenue from Tensor Processing Unit sales, reaching this milestone in the quarter that closed in June2

2

. Amazon's Trainium chips have been gaining traction with customers, including a 10-year arrangement with Anthropic to use one million Amazon Web Services' Trainium chips2

2

. The Maia line is widely seen as less mature than its rivals' hardware, making the 300,000-unit order for Maia 300 Microsoft's attempt to close that gap before cloud customers start choosing Amazon or Google specifically because of their custom silicon1

1

.

Microsoft introduced its first Maia AI chip in November 2023, followed by the Maia 200 in January 20262

2

5

5

. The Maia 200, manufactured by TSMC using 3-nanometer technology, contains over 140 billion transistors and delivers over 10 petaFLOPS of FP4 performance5

5

. Microsoft claims the Maia 200 offers more than 30% better performance per dollar than the latest-generation hardware previously deployed in its fleet5

5

. Microsoft data centers in Iowa and Arizona are already running Maia 200, with millions of servers throughout its fleet now employing its own networking, security, and virtualization silicon5

5

.

Strategic Imperative to Compete With Rivals and Reduce Costs

The urgency behind Microsoft's push into in-house AI chips reflects broader competitive pressure across the cloud industry. All three cloud giants are chasing the same prize: AI accelerators that cost less and run more efficiently than Nvidia's GPUs1

1

. Nvidia's chips remain the industry standard, but their high prices and chronic short supply have turned custom silicon from a nice-to-have into a strategic necessity1

1

. Microsoft recently told employees to stop "tokenmaxxing" and set division-level AI budgets, signaling that compute costs are becoming a constraint even at a company spending $35.8 billion per quarter on infrastructure1

1

.

At Microsoft's scale, even small savings in the cost of processing AI requests can lead to significant financial benefits5

5

. The company doesn't need Maia to replace Nvidia everywhere—it just needs its own chips to be good enough, cheap enough, and plentiful enough to transfer a meaningful share of Azure's AI workloads away from outside providers5

5

. This represents a larger threat for Nvidia investors: Microsoft's biggest clients don't need to quit buying Nvidia chips, they merely need to start needing fewer of them5

5

.

Execution Challenges and the Path to Validation

The challenge ahead centers on execution and market validation. TSMC's advanced packaging capacity is tight through 2027, and component shortages could constrain production targets1

1

3

3

. Microsoft needs to persuade enterprise customers that Maia can match or beat Nvidia on the AI workloads that matter most1

1

. The company was in early talks to supply its Maia 200 chips to Anthropic for use in the AI startup's Claude models, though no deal had been finalized at that time2

2

.

Securing Anthropic or another frontier AI startup as a major cloud customer for Maia 300 would provide critical validation that Microsoft's custom accelerator is competitive not only within its own infrastructure but also as a product in Azure's broader cloud ecosystem5

5

. Microsoft has faced delays on other custom chip projects before, with a previous next-generation Maia project in 2025 slipping at least six months because of design changes and staffing issues5

5

. For now, 300,000 chips is a bet that Microsoft's custom silicon will be ready for production scale by 20271

1

.

Market Response and Analyst Outlook

Investor optimism around Microsoft's custom silicon strategy has been evident in the stock's performance. Microsoft shares rose 1.14%, or $5.69, to $505.68 following the Maia 300 news3

3

. The stock carries a Buy rating with an average price target of $552.40, with recent analyst actions including Tigress Financial raising its forecast to $690.00, Bernstein to $647.00, and Citigroup to $600.004

4

. The stock is trading 18.6% above its 20-day SMA and 17.7% above its 200-day SMA, indicating the current run has gotten stretched versus its longer-term trend4

4

. If Microsoft's bet on cost-efficient computing through custom silicon works, Azure gains a cost advantage over Nvidia-dependent competitors. If it doesn't, Microsoft will have spent billions on a chip program that remains a generation behind its rivals1

1

.