SambaNova Launches Intelligent AI Chip SN40L, Capable of Running 50 Trillion Parameter Models


The CEO of Europe's largest AI startup, Mistral AI, stated that developing custom AI chips is a trend in the long term. With the surge in computing power demand for large models, tech giants and unicorns like OpenAI are increasingly developing their own chips or collaborating with semiconductor companies to reduce costs.
Qualcomm and ByteDance signed an AI chip supply agreement on May 26. Qualcomm will provide millions of customized ASIC chips to ByteDance to support data center AI workloads and intelligent software infrastructure. This collaboration involves not only commercial procurement but also deep integration in large-scale chip supply and manufacturing.
According to The New York Times and CNBC, SpaceX plans to invest at least $55 billion in Austin, Texas to build an AI chip manufacturing plant called "Terafab" to apply for tax incentives. The project originated from a public hearing notice in Grimes County, and if the construction phase is added, the total investment could reach $119 billion. Elon Musk first announced this ambitious plan in March.
Google is collaborating with Marvell to develop two custom AI chips, aiming to reduce its reliance on NVIDIA and strengthen its cloud service competitiveness. This move highlights the intense competition among tech giants in the computing power sector.
AI chip company Cerebras has reached a major three-year deal worth over $1 billion with OpenAI, doubling the scale of the agreement from the beginning of the year, showing OpenAI's high trust in its technology. OpenAI has committed to provide approximately $1 billion in support for Cerebras to develop data center systems and has obtained a maximum of 10% of minority equity warrants, deepening the strategic cooperation.