Moore Threads and Hygon project 135.1%-149.1% and 55.6%-70.2% revenue growth, respectively, in H1 2026, driven by China's surging demand for AI computing power amid US export controls.
Google secures $30B deal with SpaceX for xAI computing power to support Gemini Enterprise AI demand, receiving 110,000 NVIDIA GPUs from October 2023 to June 2029.
NVIDIA announced the RTX Spark, a new SOC for laptops and small desktops, competing with AMD and Qualcomm. It offers 1 petaflop of AI power, 6,144 Blackwell RTX cores, and 20 Mediatek Arm CPU cores, with power draw up to 80W.
China is building a "national computing network" to treat AI infrastructure as a public utility. Tokens, the units processed by AI, are likened to mobile data, with daily calls exceeding 140 trillion in March. This initiative is part of Beijing's "six networks" plan, with expected investment over 7 trillion yuan this year.
Technology secretary Liz Kendall has warned Britain must seize the initiative on artificial intelligence or risk being left at the mercy of technology shaped by others. Speaking about the UK's ability to develop its own capabilities, Kendall highlighted that 70% of global AI computing power is provided by just five US companies – Amazon, Google, Meta, Microsoft, and Oracle. She announced a state AI investment fund launched this month and revealed plans to become more influential in designing and manufacturing AI chips. The speech comes amid concerns the UK is struggling to develop its own path in AI despite producing academic talent and hosting companies like Google DeepMind.
China's Ministry of Industry and Information Technology reports the country has reached 1,882 exaflops in AI computing power—over 6,000 times higher than recorded in the Top500 global supercomputer ranking. When adjusted to Top500 standards, China's capacity falls to 120-230 exaflops, still far above publicly known levels. China has stopped submitting its most powerful systems to global rankings amid US export controls and geopolitical tensions.
