AWS is considering selling its custom AI chips, Trainium, to other companies for data center use. This move could pose a significant challenge to Nvidia's current AI chip market dominance, with potential annual revenues estimated at $50 billion.
An analysis in Frankfurter Allgemeine Zeitung highlights Taiwan's indispensable role in AI chip production, creating a deep global economic dependency. This reliance poses significant risks due to Taiwan's geopolitical situation and vulnerability to natural disasters.
Huawei is considering deploying its new Ascend AI chips in its cloud and AI services across Latin America, according to Mark Chen, president of Huawei Cloud Latin America. This move could increase Chinese hardware presence in a region traditionally served by US suppliers.
South Korean AI chips have secured over $30 million in export contracts with countries including the UK, Taiwan, and Vietnam. They have also passed the 'K-Perf' performance evaluation, boosting confidence for commercialization and expanding AI service applications.
A report analyzing Chinese procurement records reveals the Chinese military has sought Nvidia AI chips since 2019, even after US restrictions. Universities supporting the military have also expressed interest in high-end Nvidia processors.
Senators Elizabeth Warren and Andy Kim criticized the Trump administration for potentially allowing advanced AI chips to be exported to overseas units of Chinese firms, urging Commerce Secretary Howard Lutnick to testify before Congress.
Nvidia has launched its RTX Spark AI chip, designed to run AI agents locally on laptops and desktops, potentially replacing mouse and keyboard interactions. The move pits Nvidia against rivals like Intel, Apple, Qualcomm, and AMD in the consumer PC market.
China includes AI chips in 'secure and reliable' tech assessments for the first time, part of its Xinchuang initiative to replace foreign hardware/software with domestic alternatives, targeting AI infrastructure amid US export controls.
Microsoft is reportedly in talks to supply its custom AI chips, the Maia, to Anthropic, a move that would bolster Microsoft's position against rivals Amazon and Google in the AI silicon market. Anthropic has faced compute challenges and has existing deals with Amazon and Google.
Nvidia CEO Jensen Huang anticipates China will eventually allow imports of advanced US AI chips. While H200 chips have export licenses, China's government will decide on market access, balancing local industry protection with AI expansion. Huang noted discussions occurred during Trump's China visit.
Russian developer Baikal Electronics plans to launch two AI chips by 2029-2030, compatible with Nvidia software. The Baikal-AI-E1000 will target edge devices, and the Baikal-AI-D1000 will be a discrete GPU for servers.
Cerebras Systems' massive IPO on Thursday, with a market cap near $100 billion, highlights extreme demand for AI chips. The company offers large, custom ASICs as an alternative to Nvidia's GPUs, particularly for AI inference, and faces competition from Groq, SambaNova, and D-Matrix.
The US Commerce Department has permitted 10 Chinese firms, including Alibaba and TikTok's parent company ByteDance, to purchase NVIDIA's H200 AI processors. However, deliveries have not yet occurred, with China reportedly halting purchases due to unclear US policy changes and concerns about hidden vulnerabilities in the chips.
SoftBank has provided more than $450 million in fresh funding to Graphcore, the AI chip company it acquired in 2024. The investment forms part of a larger expected commitment this year and supports SoftBank’s push into artificial general intelligence and AI infrastructure.
Elon Musk's plans to get into the AI chip manufacturing business are going to be costly. As the New York Times and CNBC report, SpaceX is planning to invest at least $55 billion into its "Terafab" chip plant in Austin, Texas. That's according to the details of a public hearing notice filed in Grimes County, […]
Micron has emerged as one of the most valuable U.S. tech companies thanks to insatiable demand for memory that's needed in AI chips.
Anthropic is reportedly in discussions to procure high-performance AI chips from Fractile, a London-based start-up. The agreement aims to diversify the hardware supply chain of the Claude creator, namely away from Nvidia.
Alphabet's Google Cloud has surged ahead in the AI race, reporting impressive revenue growth. This performance outshines competitors Microsoft Azure and Amazon AWS. Google's unique 'full-stack' approach, including custom AI chips and advanced models, is attracting major clients. The tech giants are collectively investing over $700 billion in AI infrastructure this year.
Chinese chip designers Cambricon Technologies and MetaX Integrated Circuits reported substantial growth in the first quarter, as the artificial intelligence boom, lingering US export-control risks and China’s self-sufficiency push drove the country’s demand for home-grown computing power. Cambricon’s first-quarter revenue jumped 160 per cent from a year earlier to 2.89 billion yuan (US$423 million), while profit soared 185 per cent to 1 billion yuan, the company said in a filing on...
Chinese chip designers Cambricon Technologies and MetaX Integrated Circuits reported substantial growth in the first quarter, as the artificial intelligence boom, lingering US export-control risks and China’s self-sufficiency push drove the country’s demand for home-grown computing power. Cambricon’s first-quarter revenue jumped 160 per cent from a year earlier to 2.89 billion yuan (US$423 million), while profit soared 185 per cent to 1 billion yuan, the company said in a filing on...
Facebook parent company Meta has agreed to a multi-year deal to use Amazon’s AI chips in a deal that AWS says is worth billions of dollars.
Stocks rallied Wednesday as the S&P 500 and Nasdaq gained nearly 1% and 1.3% respectively, despite rising oil prices, with investors finding comfort in Trump's ceasefire extension. Alphabet announced its eighth-generation TPUs (TPU 8t for training and TPU 8i for inference) at Google Cloud Next, with the TPU 8t delivering nearly three times the compute performance per pod. The Club is adding BWX Technologies to its speculative Bullpen watch list, citing AI data center power demands and the company's role in nuclear component manufacturing.
Google's newest TPUs are faster and cheaper than the previous versions. But the company is still embracing Nvidia in its cloud — for now.
Google announced its eighth-generation Tensor Processing Units, splitting the lineup into TPU8t for training and TPU8i for inference. The training variant offers 121 FP4 EFlops per pod with 9,600 chips, nearly triple Ironwood's capacity, while claiming twice the performance per watt. The inference variant features 384MB of on-chip SRAM and runs in pods of 1,152 chips. Google positions this as a response to the 'agent era' requiring specialized hardware, directly competing with Nvidia's AI accelerators.
