Microsoft unveils third-generation AI Maia chip 300, planned for production in 2027 year
BIBIBI
AT A GLANCE
Microsoft plans to expand its deployment of self-developed Maia chips. Maia 300 is expected to enter production in 2027 to reduce its reliance on NVIDIA’s GPU.
Article
Microsoft’s share price reached nearly 1 a year high, as the company announced plans to deploy its own artificial intelligence chips on a large scale
BlockBeats reports that,October 12, according to BIT(Bit.com) market data, Microsoft’s share price reached nearly 1 a year high, rising 0.8%.
On the news front, Microsoft plans to gradually increase the proportion of its self-developed AI chips used in data centers to reduce its reliance on NVIDIA’s GPU. Microsoft said it will primarily rely on its self-developed Maia chip architecture to process AI model requests in the future, and plans to deploy gigawatt-scale computing capacity on the Maia series of chips.
On Monday, Microsoft announced the third-generation AI accelerator chip Maia 300, which is expected to enter production in 2027. Each data center rack can accommodate 96 Maia 300 chips, twice as many as the previous-generation Maia 200. Compared with GPU, its per-rack AI Token throughput can reach up to 7 times, while the number of Tokens processed per megawatt increases by approximately 40%.
Microsoft said that Maia 300 could improve AI service performance and reduce data center operating costs. Evercore ISI analysts estimate that, by the first quarter of 2027, Maia chips will account for approximately AI of Microsoft’s 10%chip spending.
Original link https://m.theblockbeats.info/flash/371421
Key points
01
According to BIT(Bit.com) market data, Microsoft’s share price reached nearly 1 a year high, rising 0.8%。
02
Microsoft plans to gradually increase the use of its self-developed AI chips in data centers to reduce its reliance on NVIDIA’s GPU.
03
Microsoft said it will primarily rely on its self-developed Maia chip architecture to process AI model requests in the future, and plans to deploy gigawatt-scale computing capacity on the Maia series of chips.
04
On Monday, Microsoft announced the third-generation AI accelerator chip Maia 300, expected to enter production in 2027.
05
Each data center rack can accommodate 96 Maia 300 chips, twice as many as the previous-generation Maia 200.
06
The original article says that, compared with GPU, Maia 300 can process up to AI Tokens per rack, while the number of Tokens processed per megawatt increases by approximately 7 40%。
07
Microsoft said Maia 300 is expected to improve AI service performance and reduce data center operating costs.
08
Evercore ISI analysts estimate that, by the first quarter of 2027, Maia chips will account for approximately AI of Microsoft’s 10%。
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
Microsoft hopes to use more of its self-designed Maia chips in data centers to process AI requests. Maia 300 is its third-generation AI accelerator chip. The report lists targets such as the number of chips per rack, Token throughput, and processing capacity per unit of power, but production, performance, and cost outcomes remain future plans or forecasts by relevant parties.
Why it matters to readers
If the plan is implemented, Microsoft’s use of AI computing capacity may change, potentially reducing some reliance on NVIDIA’s GPU. The processing-capacity and cost targets reported also relate to Microsoft’s ability to provide AI services more efficiently.
Beginners can watch whether Maia 300 enters production on schedule, whether its actual performance reaches the reported figures, and whether Microsoft’s use of self-developed chips actually increases; do not make judgments based solely on the share-price increase or future performance targets.
Risks and unknowns
The evidence consists of a BlockBeats report, which does not provide Microsoft’s original announcement for verification.
“October 12” and “Monday” are not explicitly tied to a year.
Maia 300 is expected to enter production in 2027, but the actual production date and deployment scale have not been confirmed by evidence.
up to 7 times the Token throughput, the approximately 40%increase in processing per unit of power, and the reduction in operating costs all lack testing conditions and actual deployment data.
Related Developments
Loading event timeline…
Maia chips account for approximately AI of Microsoft’s 10%chip spending ISI is an Evercore
The evidence does not specify the actual quantity or proportion by which Microsoft will reduce its future purchases from NVIDIA of GPU.markdown
Related concepts
AI accelerator chip
This term is not in the glossary yet. Browse related concepts in the glossary.
