Close Menu
    X (Twitter) LinkedIn
    CapitalAI DailyCapitalAI Daily
    X (Twitter) LinkedIn
    • Markets & Investments
    • Big Tech & AI
    • AI & Cybercrime
    • Jobs & AI
    • Banks
    • Crypto
    Sunday, January 11
    CapitalAI DailyCapitalAI Daily
    Home»Big Tech & AI»China Falling Behind AI Arms Race After Missing Access to US Blackwell GPUs, Says Atreides CIO Gavin Baker

    China Falling Behind AI Arms Race After Missing Access to US Blackwell GPUs, Says Atreides CIO Gavin Baker

    By Henry KanapiNovember 20, 20252 Mins Read
    Share
    Twitter LinkedIn

    Atreides Management chief investment officer Gavin Baker says the United States is pulling ahead in the global artificial intelligence hardware race, fueled by the rollout of Nvidia’s Blackwell architecture.

    In a new commentary on X, Baker says China’s semiconductor ecosystem is losing critical ground as US firms ramp the next generation of compute.

    “Blackwell will likely significantly increase the gap between the American frontier models and Chinese open source models.”

    He says domestic Chinese AI chips are no longer close enough to compete on performance. He says the gap is widening faster than analysts appreciate.

    “The domestic Chinese semiconductors are much further behind Blackwell relative to their performance vs. Hopper a year ago. This will only increase the barriers to entry for frontier models as new entrants will have to rely on increasingly outdated Chinese open source models as they attempt to catch up.”

    Baker says China is now on the back foot after missing a window of access to US hardware, noting that the strategic gap will grow even more as US defense spending fuels domestic resource and manufacturing capacity.

    “I think China will really regret not leaning into Trump’s willingness to sell them the ‘B30.’ Blackwell will alter the relative leverage of the US vs. China, especially as the truly immense DoD effort to increase domestic rare earth mining and refining pays off over the next two years. I think the technological solutions to refining that are being pursued are underappreciated and at least some are likely to succeed.”

    The Nvidia Blackwell B200 GPU is a significant leap from the Hopper-architecture H100 GPU. For AI inference, it delivers up to 30 times faster performance than the H100 system, specifically for trillion-parameter large language models (LLMs), and is up to four times faster than the H100 for LLM training. It also offers up to 25 times better energy efficiency for inference workloads and comes with 8 TB/s of memory bandwidth, which is approximately 2.4 times faster than the H100’s 3.35 TB/s.

    Disclaimer: Opinions expressed at CapitalAI Daily are not investment advice. Investors should do their own due diligence before making any decisions involving securities, cryptocurrencies, or digital assets. Your transfers and trades are at your own risk, and any losses you may incur are your responsibility. CapitalAI Daily does not recommend the buying or selling of any assets, nor is CapitalAI Daily an investment advisor. See our Editorial Standards and Terms of Use.

    Blackwell China Gavin Baker Nvidia US
    Previous ArticleGoldman Sachs Says AI Not Yet in Bubble Based on 175-Year of Tech Cycles – ‘Path Will Be Up and to the Right’
    Next Article Jensen Huang Destroys AI Bubble Fears, Says Compute Capacity Substantially Less To Support Global Demand

    Read More

    AI Macro Guru Reveals How to Chain Grok and Gemini With Multiple LLMs To Create 1,000x Powerful Prompts

    January 11, 2026

    China’s Top AI Leaders Say Odds of Beating the US in Artificial Intelligence Race Are Below 20%: Report

    January 11, 2026

    Anthropic Co-Founder Warns Society Is Not Ready for Self-Improving AI, Says Early Warning Signs Are Appearing

    January 11, 2026

    Billionaire Chamath Palihapitiya Says SpaceX Will Never IPO and Could Reverse Merge Into Tesla

    January 11, 2026

    Michael Burry Calls for $1,000,000,000,000 Nuclear Buildout To Power America and Keep Up With China

    January 11, 2026

    Bank of America Warns Market Is Missing the Real AI Moat, Says Investors Should Stick With Big Players Partnering With Data Centers

    January 10, 2026
    X (Twitter) LinkedIn
    • About
    • Author
    • Editorial Standards
    • Contact Us
    • Privacy Policy
    • Terms of Service
    • Cookie Policy
    © 2025 CapitalAI Daily. All Rights Reserved.

    Type above and press Enter to search. Press Esc to cancel.