Tenstorrent And LG Electronics Partner To Build AI Chips For Smart TVs And Automotive Products

JAKARTA - Canadian AI computer design startup Tenstorrent announced on Tuesday, May 30 that it will partner with South Korean consumer electronics company LG Electronics Inc to build chips that power smart TVs, automotive products, and data centers.

"Tenstorrent, founded in 2016, designs computers to train and run artificial intelligence models, as well as work on software and hardware," CEO Jim Keller said in an interview, quoted by Reuters. Keller was an engineer known for his pioneering work in designing chips at Apple Inc, Tesla Inc, and chipmaker Advanced Micro Devices Inc (AMD).

Keller, who was one of the early investors in Tenstorrent, will take over as chairman of the company in 2023. According to data from research firm PitchBook, the current value of Tenstorrent is US$1 billion (IDR 15 trillion), however, until now there has been no disclosure regarding customers. they.

According to David Bennett, Chief Customer Officer of Tenstorrent, at first, LG will use Tenstorrent's AI chip blueprints to design their own chips, but this partnership is of a more strategic nature.

"On top of that, we're also looking at the technology that LG is developing. Whether it can't be something we use in our own products or maybe with other future customers," said Bennett.

Tenstorrent has also designed a processor chip using the relatively new open standard chip architecture, RISC-V, which competes with Arm Ltd.'s architecture. While many chip startups focus on one type of chip, Keller said his team is developing both types of AI chips and processors because they need to work closely together to deal with rapidly changing AI models.

"We have to look at the whole picture. ... It's still pretty early days. And it's building on the components that are available," Keller said of current developments in AI and AI hardware.

"In the last five years, people have learned a lot about how this thing works and made real progress. However, it seems we are still a long way from 'this is the right way to do it, the best way to do it, or the final thing," Keller added.