DoubleAI Launches CEM AI Category and Maxwell Cost-Cutting Tool
Maxwell will be broadly available from late October, with DoubleAI charging customers a share of the computing costs they save.
DoubleAI on October 19 unveiled what it calls a new AI category, the Code Execution Model (CEM), alongside its first product, Maxwell, which the company says is designed to lift GPU and CPU performance and cut corporate computing costs.
The company does not charge a flat license fee. Instead, customers pay a portion of the costs they save, according to DoubleAI. One early customer cut its computing costs by 67%, the company said.
AI company LTX doubled its model inference speed using Maxwell, according to the report. DoubleAI said Maxwell will be broadly supplied starting in late October. The company is led by CEO Amnon Shashua, a co-founder of the autonomous driving company Mobileye.
Funding and hardware
DoubleAI has raised $220 million from Lightspeed Venture Partners, Bessemer Venture Partners, Nvidia and Dell, and used about half of that funding to buy 1,000 Nvidia B300 chips.
The pitch lands as chip rental costs climb. According to SemiAnalysis, renting an Nvidia H100 on a one-year contract cost 60% more per hour over the past year.
Quick answers
When will Maxwell be available?
DoubleAI said Maxwell will be broadly supplied starting in late October.
How does DoubleAI charge for Maxwell?
Customers pay a fee based on a portion of the computing costs they save, according to DoubleAI.