Microsoft unveils Maia 200 chip as it powers OpenAI models

Microsoft recently unveiled its second-generation AI chip, the Maia 200, designed to significantly enhance AI model inference tasks. This new chip boasts impressive performance, delivering over 10 petaflops in FP4 and 5 petaflops in FP8. Microsoft claims the Maia 200 is three times faster than Amazon's Trainium chips in FP4 performance and outperforms Google's latest TPU in FP8, offering Azure users a potential 30% cost saving.

The Maia 200 is built on TSMC's 3-nanometer technology, featuring over 140 billion transistors and 216GB of HBM3e memory. Microsoft is already deploying it in its U.S. data centers, including Iowa, Arizona, and the U.S. Central region. Internally, the chip powers Microsoft Foundry, Microsoft 365 Copilot AI, and supports OpenAI's GPT-5.2 models. Microsoft also provides a Maia SDK for developers, with plans for wider customer availability through Azure.

Beyond hardware, IBM is updating its Partner Plus program to accelerate AI growth for its partners. The enhancements include better incentives, tools, and co-marketing support, offering an extra 15 points on sales and recognition for multiyear contracts. IBM aims to unlock a $200 billion opportunity for partners in the midmarket, integrating new AI-driven selling experiences and support for hyperscaler marketplaces.

In the startup world, Ricursive Intelligence, less than a year old, has already raised $335 million at a $4 billion valuation. This Palo Alto-based company, founded by Anna Goldie and Azalia Mirhoseini, focuses on creating AI that can improve itself autonomously. Meanwhile, Hammerspace promoted Tony Asaro to Chief Business Officer to lead global sales, addressing the rising demand for AI infrastructure and

Key Takeaways

  • Microsoft launched its Maia 200 AI chip, which is three times faster than Amazon's Trainium in FP4 and outperforms Google's latest TPU in FP8.
  • The Maia 200 chip, built on TSMC's 3nm process with over 140 billion transistors, offers 30% higher performance per dollar.
  • Microsoft is deploying the Maia 200 in its U.S. data centers and uses it to power Microsoft 365 Copilot AI and OpenAI's GPT-5.2 models.
  • IBM is enhancing its Partner Plus program with new incentives and AI-driven tools, aiming for a $200 billion opportunity for partners in the midmarket.
  • Startup Ricursive Intelligence, founded by Anna Goldie and Azalia Mirhoseini, raised $335 million and is valued at $4 billion for its work on self-improving AI.
  • Hammerspace appointed Tony Asaro as Chief Business Officer to address demand for AI infrastructure and data strategies, expanding its reseller channel to nearly 200 partners.
  • The UK government plans to allow AI systems to use public data from institutions like the Met Office and National Archives, creating a

    Microsoft launches Maia 200 AI chip for faster responses

    Microsoft introduced its second-generation AI chip, the Maia 200, to improve how AI models serve user responses. This new chip focuses on "inference" and is three times faster than Amazon's Trainium chips in FP4 performance. It also performs better than Google's latest TPU in FP8. The Maia 200 has over 100 billion transistors and is made using TSMC's 3-nanometer technology. Microsoft is already using it in data centers in Iowa and Arizona and offers a software development kit for developers.

    Microsoft's Maia 200 AI chip outperforms rivals

    Microsoft unveiled its Maia 200 AI chip, claiming it beats offerings from AWS and Google. This chip provides three times the performance of AWS and can save Azure users 30% on costs. Built on TSMC's 3nm process, it features over 140 billion transistors and 216GB of HBM3e memory. The Maia 200 is designed for AI inference and also helps with creating synthetic data and reinforcement learning. Microsoft has launched a preview of the Maia SDK for developers.

    Microsoft Maia 200 chip triples competitor AI power

    Microsoft announced its new Maia 200 accelerator chip, built for AI inference tasks. Scott Guthrie, a Microsoft executive, stated the chip delivers over 10 petaflops in FP4 performance and 5 petaflops in FP8. He also said it is three times faster than Amazon's Trainium in FP4 and better than Google's latest TPU in FP8. The Maia 200 is already deployed in Microsoft's U.S. central data center and will power Microsoft Foundry and 365 Copilot AI. Microsoft plans wider customer availability through Azure.

    Microsoft strengthens cloud with new Maia 200 AI chip

    Microsoft introduced its second-generation AI chip, the Maia 200, to boost its cloud services. This chip offers 30% higher performance for the same price compared to other options. Microsoft is deploying the Maia 200 in its U.S. data centers, starting with the U.S. Central region and then U.S. West 3. The chip will be used by Microsoft's internal AI team and for Microsoft 365 Copilot, with plans for wider customer access. Scott Guthrie noted that the Maia 200 is made using TSMC's 3-nanometer process and can connect up to 6,144 chips.

    Microsoft's Maia 200 AI accelerator boosts inference

    Microsoft launched Maia 200, a powerful AI chip engineered to improve AI inference tasks. Built on TSMC's 3nm process, it offers over 10 petaFLOPS in FP4 performance and 5 petaFLOPS in FP8. Microsoft states the Maia 200 is three times faster than Amazon's Trainium in FP4 and outperforms Google's latest TPU in FP8. This chip is already in Microsoft data centers in Iowa and Arizona, supporting OpenAI's GPT-5.2 models and Microsoft 365 Copilot. Microsoft also provides a Maia SDK for developers to build and optimize AI models.

    Microsoft unveils Maia 200 AI chip for better efficiency

    Microsoft unveiled its Maia 200 AI chip, specifically designed to make AI inference workloads more efficient. This chip, built on TSMC's 3-nanometer process with over 140 billion transistors, delivers impressive performance, including 10 petaFLOPS in FP4. Microsoft claims it is three times faster than Amazon's Trainium and outperforms Google's TPU. The Maia 200 also offers 30% better performance per dollar. It is currently deployed in Microsoft's US Central data center and will power services like OpenAI's GPT-5.2 models and Microsoft 365 Copilot.

    IBM boosts Partner Plus program for AI growth

    IBM is updating its Partner Plus program to help partners succeed more with AI. David La Rose, IBM's general manager of partner ecosystem sales, stated that the changes will offer better incentives, tools, and co-marketing support. Partners will get an extra 15 points on sales and recognition for multiyear contracts, aiming for steady recurring revenue. Kareem Yusuf, IBM's senior vice president of ecosystem, mentioned new AI-driven selling experiences and support for hyperscaler marketplaces. IBM sees a $200 billion opportunity for partners in the midmarket.

    IBM enhances Partner Plus for AI success

    IBM is speeding up its Partner Plus Program to help partners grow with AI. Kareem Yusuf, IBM's Senior Vice President of Ecosystem, announced new incentives, more benefits, and AI-driven selling tools. The program will also offer scalable co-marketing funds and easier ways to sell through hyperscaler marketplaces and IBM Agent Connect. IBM is also improving how its sellers and partners work together to close deals faster. The company aims to remove obstacles for resellers and help service partners and software vendors use IBM technology more effectively.

    Startups aim for AI that improves itself

    Two new startups, including Ricursive Intelligence, are working to create AI that can improve itself without human help. Ricursive Intelligence, based in Palo Alto, California, was founded by Anna Goldie and Azalia Mirhoseini. They believe that if AI systems can design better chips, those chips will then create even better AI systems in a continuous loop. Ricursive Intelligence, though less than a year old with fewer than 10 employees, has already raised $335 million and is valued at $4 billion.

    Hammerspace names Tony Asaro Chief Business Officer

    Hammerspace promoted Tony Asaro to Chief Business Officer to lead its global sales and business development. Asaro will focus on meeting the growing demand from governments and businesses for AI infrastructure and data strategies. Hammerspace aims to provide "AI Anywhere" solutions, allowing AI projects to run where compute and models exist without sacrificing speed or control. The company has strong sales momentum, growing its reseller channel to nearly 200 partners. Hammerspace also expanded internationally, opening an Asia headquarters in Singapore and increasing presence in other regions.

    CRN hosts AI Security Week 2026

    CRN is holding its first AI Security Week in 2026 to explore how AI is changing cybersecurity. The event will cover important trends and opportunities for companies that provide security solutions and services. It will look at essential controls for AI security and how top vendors use "agentic AI" to improve Security Operations Centers. The week will also discuss how AI chips are making security better and how startups are using AI agents for cyber defense. Secure AI use needs good visibility, strong user verification, and tools to prevent data loss.

    UK plans for AI to use public data

    The UK government plans to allow AI systems to use data from public institutions like the Met Office and National Archives. Ministers will fund researchers to test how legal data from the National Archives can help small businesses get quick answers. The government also plans to license content from institutions such as the National History Museum and National Library of Scotland for AI development. This initiative, called a "creative content exchange," will create a marketplace for digital cultural assets. A pilot platform is expected to launch in the summer.

    AI reshapes society and economy not just a tool

    Fred Cavazza's analysis suggests that AI is not just a tool but a force reshaping society and the economy. He argues that AI is a major driver of productivity, disrupting economic and social balance and transforming office jobs. AI is creating a divide between workers who use it and those who do not. Cavazza believes AI marks our entry into the fourth industrial revolution by letting machines handle human thinking and creativity. He emphasizes the need for new legal and tax rules for AI agents and public debate on its wide-ranging impacts.

    AI writing tools may standardize our thoughts

    AI writing tools might be making our thoughts and language too similar, creating a "linguistic flattening." The article explains that language shapes our thoughts, and as AI text generators become common, they risk standardizing how we think and express ourselves. It warns that Western ways of thinking about conflict and storytelling could dominate, changing cultural metaphors and proverbs. The author suggests using "hybrid intelligence" to combine human intuition with AI's analytical power. We must be aware and actively use AI to avoid simply accepting its default settings.

    Boston University launches online AI business master's

    Boston University's Questrom School of Business launched a new Online MS in Artificial Intelligence in Business. This master's degree is for leaders who want to use AI responsibly and effectively in their organizations. Paul Carlile, a Senior Associate Dean, stated the program focuses on redesigning workflows and governance for AI. The program is 100% online, consists of 32 credits, and costs $25,000. The first group of students will begin in August 2026.

    Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

Microsoft Maia 200 AI Chip AI Inference Data Centers Azure Microsoft 365 Copilot TSMC 3nm Process AI SDK IBM Partner Program AI Ecosystem AI Growth Self-Improving AI AI Infrastructure AI Security Cybersecurity Agentic AI Public Data for AI AI Development Societal Impact of AI Economic Impact of AI Workforce Transformation AI Ethics AI in Business Education AI Writing Tools Linguistic Flattening Hybrid Intelligence AI Governance Cloud Services OpenAI

Comments

Loading...