成人小说亚洲一区二区三区,亚洲国产精品一区二区三区,国产精品成人精品久久久,久久综合一区二区三区,精品无码av一区二区,国产一级a毛一级a看免费视频,欧洲uv免费在线区一二区,亚洲国产欧美中日韩成人综合视频,国产熟女一区二区三区五月婷小说,亚洲一区波多野结衣在线

Amazon.com Inc. founder Jeff Bezos joined Samsung in a $700 million bet on Tenstorrent, valuing the AI chip startup with ambitions of taking on Nvidia Corp. at about $2.6 billion.

Tenstorrent, which hopes to create a chip to try and break Nvidia’s stranglehold on the AI business, raised capital in a funding round led by South Korea’s AFW Partners and Samsung Securities, founder and semiconductor pioneer Jim Keller said in an interview. Bezos Expeditions joined LG Electronics Inc. and Fidelity in that financing, betting on Keller’s pedigree and the booming opportunity in artificial intelligence tech.

The money will be used to build out Tenstorrent’s engineering team, invest in its global supply chain and build large artificial intelligence training servers to help demonstrate its technology.

As the quest for more power and cost efficiency in AI ramps up, smaller companies are sprouting up trying to snatch market share from Nvidia’s power-hungry chips. Tenstorrent, an Nvidia neighbor in Santa Clara, California, is one of many now engineering solutions aimed at delivering a more affordable path to AI development. That’s built on open-source and commonplace technology, avoiding complex and pricey components like the high-bandwidth memory Nvidia favors.

“You can’t beat Nvidia if you use HBM, because Nvidia buys the most HBM and has a cost advantage,” Keller said. “But they’ll never be able to bring the price down the way HBM is built into their products and their sockets.”

Nvidia offers developers a full suite of proprietary technology, covering everything from the chips to the interconnects and even data center layouts, with the promise of all parts working better because they were designed in concert. Companies like rival Advanced Micro Devices Inc. and Tenstorrent are instead aiming for greater interoperability with other technology providers, whether through shared industry standards or opening their designs for others to use.

Tenstorrent is also a proponent of an alternative kind of logic processor based on an open standard called RISC-V, which poses a challenge to Arm Holdings Plc. Keller, known for his silicon design work at Apple Inc., Tesla Inc. and AMD, is an advocate.

“In the past, I worked with proprietary tech and it was really tough,” Keller said. “Open source helps you build a bigger platform. It attracts engineers. And yes, it’s a little bit of a passion project.”

Much like RISC-V and Japanese partner Rapidus Corp., Tenstorrent still has a lot to prove. To date, the nascent company has signed contracts with customers totaling nearly $150 million, which pales in comparison to Nvidia’s tens of billions of dollars of datacenter revenue each quarter.

Tenstorrent plans to release a new AI processor at a cadence of every two years, Keller said. Nvidia, on the other hand, intends to refresh its AI chip offerings on an annual cycle, its boss Jensen Huang said in June.

AFW Partners invested after hearing positive feedback from Korean companies already collaborating with Tenstorrent, such as LG, said managing director Bonil Koo.

Tenstorrent’s first chips were made by GlobalFoundries Inc., and the next iterations will come from Taiwan Semiconductor Manufacturing Co. and Samsung Electronics Co., the company said. It’s begun designing for cutting-edge 2-nanometer fabrication as well. TSMC and Samsung will commence their mass production at that scale next year, and Tenstorrent is in discussions with them and Japan’s Rapidus, which aims for 2nm output in 2027.

Other investors in this funding round include Export Development Canada, Healthcare of Ontario Pension Plan, [hotlink]Hyundai Motor[/hotlink] Group and Baillie Gifford.