Google Cloud is giving Y Combinator startups access to a dedicated, subsidized cluster of Nvidia GPUs and Google Tensor Processors to construct AI models. This is a component of Google Cloud's effort to ingratiate itself with promising early-stage AI startups within the hopes that a few of them will grow into huge, compute-hungry firms.
“We wish to surround them with a number of love and heat early of their lifecycle so that they can get comfortable with constructing and dealing on the Google Cloud platform,” said James Lee, Google Cloud's general manager of startups and AI business, in an interview with TechCrunch. “If they stick around, we grow with them and change into their partner throughout their lifecycle.”
Specifically, Google will provide a dedicated cluster with priority access for the YC startups in summer 2024, together with $350,000 in cloud credits over two years for every participating startup. The idea is that a future unicorn will fall in love with constructing on Google Cloud. Five percent of Y Combinator startups have change into unicorns with valuations of over a billion dollars over the past 18 years, YC group partner Diana Hu tells TechCrunch, so it's a superb bet for Google on this partnership.
“It's very exciting that the following generation of startups are prone to be not only unicorns, but decahorns, they usually're being built immediately,” Hu said. “Cloud providers still need to catch up by way of pricing, but they know that in the event that they catch them early, they'll ride the wave with them.”
On the opposite hand, Y Combinator says it might probably get more AI startups to use by offering generous computing resources alongside its investments and advice. For early-stage AI startups, Hu says one of the crucial common problems she hears is that the startups have limited computing capability. Large firms can sign multi-year, large contracts with cloud providers for GPU access, but small startups often get left behind. She says partnerships like this are very helpful, especially if you have got a dedicated cluster—this is particularly necessary for training AI models.
“GPU and AI workloads are more like high-performance computing workloads in batches. The server doesn't have to be running on a regular basis, but when the workload is peaking, you would like a number of them,” Hu said. “That's why we’ve got a dedicated cluster that YC firms can easily use.”
As a part of the partnership, Google can also be offering YC startups $12,000 in Enhanced Support credits and a free yr of Google Workspace Business Plus. YC startups also can connect with Google's in-house AI experts during monthly office hours.
Startup accelerators and enterprise capital firms are increasingly offering GPU clusters lately. Andreessen Horowitz has reportedly a stock of 20,000 GPUs to draw AI startups. Google Cloud and Y Combinator declined to reveal the precise amount of the deal, but Hu says it’s enough for YC's startup firms to coach models.
