In the age of AI, companies are in an intense race to expand their computing power, but what if they could do more with what they already have?
That’s the promise of a new partnership between Nvidia and Spectro Cloud, a San Jose-based startup valued at around $750 million and backed by Goldman Sachs Alternatives.
Spectro Cloud announced Tuesday it was launching a new platform, PaletteAI, with Nvidia in order to help companies manage their AI infrastructure and tackle a growing problem in the AI boom: wasted computing power.
Nvidia’s products, from hardware like chips to software, are widely used in modern AI infrastructure, but managing those systems efficiently can be complicated.
“AI infrastructure is super expensive,” Tenry Fu, CEO and cofounder of Spectro Cloud, told Business Insider, adding that many companies’ GPU utilization, or the percentage of processing power that’s actually being used at a given time, is only around 30%.
“That’s not really the best way to put such an expensive hardware into use,” he said, adding that using PaletteAI can raise GPU utilization to 60%, effectively doubling the efficiency.
Optimizing GPU usage can save money and energy, as well as speed up AI development. It also helps companies get the most bang for their buck when it comes to the infrastructure they’ve already invested in.
The platform, which is set to debut this week at Nvidia’s GTC conference, is integrating with Nvidia’s AI Enterprise software suite. It’s intended to simplify how companies adopt and manage AI systems in a way that is easier and more secure.
“We live in a really interesting time now where, for the first time and perhaps ever, we have — because of AI — everything changing rapidly and at the same time,” Dave Cope, Spector Cloud’s chief revenue and marketing officer, told Business Insider. He said the “tremendous amount of complexity” can stifle or slow down AI innovation and adoption.
While PaletteAI comes integrated with Nvidia products, it’s an open platform that can be used with other hardware and software vendors.
Saad Malik, CTO and cofounder, said the company provides the “glue layer” that allows all the different hardware and software components used in AI systems to work together.
“Growing adoption of AI across every industry calls for scalable, adaptable infrastructure that bridges the data center and the edge,” Anne Hecht, senior director for Enterprise AI at Nvidia, said in a statement.
Have a tip? Contact this reporter via email at kvlamis@businessinsider.com or Signal at @kelseyv.21. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.
Read the full article here
