AI Arms Race Heats Up As Startups Borrow Billions To Buy Nvidia GPUs

Amid rising infrastructure demands, AI startups like Crusoe Energy Systems are tapping large-scale credit lines, PitchBook reported. Crusoe last month raised a $750 million credit facility from Brookfield Asset Management (NYSE:BAM) to expand its data centers and acquire more graphics processing units from Nvidia Corp. (NASDAQ:NVDA).

Contents

Why Startups Are Turning to Debt Sooner Large Debt Deals Backed by GPUs Some Lenders Are Cautious Venture Debt’s Rapid Expansion

That deal reflects a broader trend in venture financing. AI and machine learning startups have secured an outsized share of venture debt so far, accounting for more than a third of the $30 billion deployed across the U.S. and Europe, according to PitchBook. That marks a significant increase from 2024, when such companies attracted 24.9%, or $22.9 billion, of total debt funding.

Don’t Miss:

Accredited investors can claim pre-IPO shares of Mode Mobile for just $0.30—with up to 120% bonus shares—before this Uber-style disruption hits the public markets
$100k+ in investable assets? Match with a fiduciary advisor for free to learn how you can maximize your retirement and save on taxes – no cost, no obligation.

Why Startups Are Turning to Debt Sooner

According to PitchBook, the shift toward debt financing is being driven by rising compute costs and high early-stage valuations in AI.

“There are companies coming out of the gate with these large chunky seed rounds,” Silicon Valley Bank Managing Director, Early-Stage Startups Bo Ren told PitchBook. “But then there’s a Series A valuation drop-off because they can’t live up to the VC expectations.”

Ren said that many AI startups are turning to venture debt earlier in their lifecycle as they struggle to maintain growth targets set by their inflated early valuations.

Ren also cited infrastructure costs as a key factor: “Given the cost of compute, it’s very much like the price of gas, it just keeps going up and up. That’s where a lot of the venture debt conversations start,” she told Pitchbook.

PitchBook data shows that the median pre-money valuation for AI startups has climbed to $25 million so far this year, up from $15 million in 2024.

Trending: Named a TIME Best Invention and Backed by 5,000+ Users, Kara’s Air-to-Water Pod Cuts Plastic and Costs — And You Can Invest At Just $6.37/Share

Large Debt Deals Backed by GPUs

Several AI infrastructure startups have recently raised major venture debt deals to purchase high-performance chips.

Lambda Labs also raised a $500 million special-purpose financing vehicle, collateralized by GPUs it already owned, to expand its chip inventory.

GPU cloud provider CoreWeave (NASDAQ:CRWE), which recently went public, raised $2.3 billion in debt with similar terms, PitchBook reported. The loan was backed by the company’s Nvidia GPU stockpile.

See Also: This AI-Powered Trading Platform Has 5,000+ Users, 27 Pending Patents, and a $43.97M Valuation — You Can Become an Investor for Just $500.25

Some Lenders Are Cautious

“How long is the useful life of those chips? If it’s 10 years or even seven years, that makes sense,” CIBC Innovation Banking Executive Managing Director and Co-Head Paul McKinlay told PitchBook. “But if it’s two years, that starts to become a tougher proposition.”

Venture Debt’s Rapid Expansion

PitchBook reported that total venture debt deployment reached $53.3 billion in 2024, a 94.5% increase from 2023. That growth was primarily fueled by later-stage startups preserving valuations amid slowing VC rounds.

In 2025, AI is driving the next wave of venture debt demand—now extending to earlier-stage companies.

Read Next: Warren Buffett once said, “If you don’t find a way to make money while you sleep, you will work until you die.” Here’s how you can earn passive income with just $10.

Image: Shutterstock

“ACTIVE INVESTORS’ SECRET WEAPON” Supercharge Your Stock Market Game with the #1 “news & everything else” trading tool: Benzinga Pro – Click here to start Your 14-Day Trial Now!

Get the latest stock analysis from Benzinga?