Microsoft will shift to its own Maia AI chips, aiming for multi‑gigawatt deployments and reducing reliance on Nvidia GPUs.
What happened
Microsoft announced that its third‑generation
AI acceleratorA specialized chip designed to speed up artificial‑intelligence tasksAI accelerator
, the Maia 300, will enter production in 2027. The company plans to pack 96 Maia 300 chips into each data‑center rack, twice the number used for the Maia 200.The Maia 300 can handle up to seven times more AI tokens per rack and offers 40% more tokens per megawatt compared with traditional GPUs. Microsoft expects to run multiple gigawatts of Maia chips across its Azure cloud, according to Executive Vice President Scott Guthrie.
Earlier versions, the Maia 100 (2023) and Maia 200 (2024), already power parts of Microsoft 365 Copilot and GitHub Copilot. Finance chief Amy Hood said using its own silicon has helped improve margins over the past decade.
Despite the shift, Microsoft will still partner with Nvidia and AMD for model training, and Nvidia chips will remain in some Surface laptops.
Why it’s important
Running its own
gigawattA unit of power equal to one billion watts, used to measure large‑scale electricity usegigawatt
‑scale AI compute could lower costs for Azure customers. Lower hardware expenses improve Microsoft’s profit margins and may make its cloud services more competitive.The move also reduces dependence on Nvidia, which currently dominates the AI‑chip market. Diversifying suppliers can protect Microsoft from supply‑chain shocks and pricing pressure.
Investors watch chip strategies because they affect both capital‑expenditure outlooks and long‑term profitability of cloud providers.
Microsoft’s Maia 300 aims to deliver up to seven times more AI tokens per rack than its predecessor, boosting efficiency dramatically.
What could change
If the Maia 300 meets performance targets, Microsoft could accelerate AI service rollout and attract more enterprise customers to Azure. Conversely, if the chips lag behind Nvidia’s roadmap, Microsoft may need to keep buying external GPUs, slowing cost savings.
Regulatory or trade restrictions on semiconductor components could also affect the timeline for large‑scale Maia deployment.
If Maia chips underperform, Microsoft’s AI cost advantage could be delayed, impacting its cloud margin outlook.
What does this mean for investors?
Retail investors may see Microsoft’s stock benefit from lower cloud operating costs and a stronger AI offering. ETFs that focus on cloud computing or AI hardware, such as CLOU or BOTZ, could also feel indirect effects.
Bond investors should monitor Microsoft’s capital‑expenditure plans, as large chip investments may influence cash flow and credit metrics.
Disclaimer: This post was created with the support of artificial intelligence and reviewed by our editorial team. It is provided for educational purposes only and does not constitute financial advice.