§ TECH · 2 MIN READ
The Professor Who Put Many Brains on One Chip
In the 1990s Kunle Olukotun argued that a computer chip should hold many small processors instead of one big one. Everyone said no. Now every phone works his way, and his AI chip company just raised 350 million dollars.
By Culture
THE SHORT VERSION
Pull out your phone. Somewhere inside it is a chip with several small processors working side by side instead of one big one doing everything. That idea has a name, the chip multiprocessor, and thirty years ago the industry thought it was wrong. A young Stanford professor named Kunle Olukotun thought it was the future. He was right, and the whole world runs on it now.
In the mid 1990s, the way to make a chip faster was to make its one processor bigger and hotter. Olukotun's Hydra project at Stanford argued the opposite: put several simpler processors on one piece of silicon and let them share the work. Simpler, cooler, and much easier to scale.
He proved it by building a company. Afara Websystems, which he co-founded, was bought by Sun Microsystems in 2002, and its design became Sun's Niagara chip, one of the first commercial multicore processors. Today a laptop with eight cores is normal. A phone with six is normal. That is Hydra, grown up.
In 2017 Olukotun did it again, this time for artificial intelligence. AI models like the ones writing emails and reading X-rays need enormous amounts of math done in a very specific pattern, and the ordinary chips that run games and spreadsheets are a clumsy fit. SambaNova, which he co-founded with Rodrigo Liang and fellow Stanford professor Chris Ré, designs chips shaped around the way AI actually computes.
In February 2026 the company raised a 350 million dollar Series E. A Series E is a late, large round, the kind you raise when you are big and getting bigger. That brought SambaNova's total known funding to about 1.5 billion dollars.
