Gruve raises $50 million to solve what its CEO calls AI's biggest problem: power

• Gruve raised $50 million to power AI inference as demand shifts from training.
• The startup stitches together unused data center power across major US cities.
• Gruve cofounder and CEO Tarun Raisoni sold two previous startups in nine-figure deals.

As the focus in AI shifts from training to inference, infrastructure startup Gruve has raised $50 million to close a widening power gap and help put AI models to work.

Gruve, which launched in 2024, partners with data center and colocation providers like Lineage and OpenColo to tap their unused power and space. The company says it now has access to roughly 500 megawatts of power across a network of data centers in major US cities.

"The biggest challenge today in AI is we don't have enough power," said Tarun Raisoni, Gruve's CEO and cofounder. "We have found the stranded power, and we are bringing the software to stitch it together."

Raisoni, a serial entrepreneur, previously founded data center startups Rahi and ZPE, which were acquired in nine-figure deals by electrical infrastructure companies Wesco and Legrand, respectively.

Gruve's Series A follow-on brings total funds raised to $87.5 million, the company said. Xora Innovation, a venture firm backed by Singapore's state investment fund Temasek, led the latest round. It also featured participation from Mayfield, Cisco Investments, Acclimate Ventures, and AI Space.

Gruve now offers 30 megawatts available to order across four sites in California, New Jersey, Texas, and Washington, the company said, with customer data running in its California and New Jersey locations.

Geographic distribution is key, Raisoni added, with software that can route requests to the nearest server location, resulting in faster transmission and lower costs.

Gruve typically works with neoclouds that supply the hardware, Raisoni told Business Insider, after which it handles setup, management, and day-to-day operations.

Unlike cloud giants, Gruve provides hands-on engineering support, as many companies lack in-house machine learning and data science talent, Raisoni said.

Gruve's customers include neoclouds, AI startups, and corporations, and most fall into that third bucket, like Bio-Rad, PayPal, Cisco, and Stanford Health Care, Raisoni said. Down the line, Gruve plans to expand in Japan and Western Europe.

Gruve has about 600 employees, 70% of whom are based in India and focus on security operations. Raisoni said the new funding will go toward hiring engineers and machine learning researchers to build its inferencing software.