Cerebras
In plain wordsAn American chip company that makes a single giant processor the size of a whole silicon wafer, used to train and run AI models very quickly.
An American AI hardware company that builds the wafer-scale enginewafer-scale engine, n.: A single giant processor made from nearly a whole dinner-plate-sized silicon wafer instead of the small chips normally cut from it, so data rarely has to leave the chip., a processor made from an entire silicon wafer rather than a single small die, with hundreds of thousands of cores and large amounts of on-chip SRAMSRAM, abbr.: The fastest kind of computer memory, built directly into processor chips. It is quick but takes up a lot of space, so chips hold only a little of it. that sidestep much of the memory wallmemory wall, n.: The growing gap between how fast chips can calculate and how fast they can fetch data from memory. Processors increasingly sit idle, waiting for numbers to arrive. faced by GPUs. Its third generation, WSE-3 (2024), has about 4 trillion transistors and 900,000 cores and is manufactured by TSMCTSMC, prop. n.: The Taiwanese company that manufactures most of the world's advanced computer chips, including nearly all of the specialized chips used to train and run AI.. First sold as CS-series systems for training, it turned in 2024 toward a cloud inferenceinference, n.: Drawing a conclusion from evidence. In today's AI it most often means running an already-trained model to get an answer, as opposed to training it. service that serves language models at very high token rates, with users including Meta's Llama API, Mistral AI and Perplexity, and in January 2026 it agreed a multiyear deal to supply computing to OpenAI. Its dependence on the Abu Dhabi firm G42 as a customer and investor delayed a planned stock listing during a US national-security review.
Origin
Founded in 2015 in Sunnyvale, California, by Andrew Feldman, its chief executive, Gary Lauterbach, Michael James, Sean Lie and Jean-Philippe Fricker, who had worked together at the server maker SeaMicro; it unveiled its first wafer-scale chip in August 2019.
Related
- wafer-scale enginewafer-scale engine, n.: A single giant processor made from nearly a whole dinner-plate-sized silicon wafer instead of the small chips normally cut from it, so data rarely has to leave the chip.
- AI acceleratorAI accelerator, n.: Any chip built or tuned to speed up AI — graphics chips, Google's TPUs, phone 'neural engines' and many startup designs. Such chips do the heavy math of training and running models.
- inferenceinference, n.: Drawing a conclusion from evidence. In today's AI it most often means running an already-trained model to get an answer, as opposed to training it.
- SRAMSRAM, abbr.: The fastest kind of computer memory, built directly into processor chips. It is quick but takes up a lot of space, so chips hold only a little of it.
- memory wallmemory wall, n.: The growing gap between how fast chips can calculate and how fast they can fetch data from memory. Processors increasingly sit idle, waiting for numbers to arrive.
- NvidiaNvidia, prop. n.: An American chip company whose graphics processors became the main hardware for training and running AI, making it one of the most valuable companies in the world.