Fintech8 May 20263 min readBy Fintech News Desk· AI-assisted

Coinbase Posts US$394M Q1 Loss As AWS Cooling Failure Knocks Exchange Offline The Same Day

Coinbase booked a US$394 million GAAP loss on Q1 2026 revenue of US$1.41 billion, down 31% year-on-year, while CEO Brian Armstrong leaned on stablecoin growth and an aggressive AI restructure to defend the result. Hours later, an AWS cooling failure in Northern Virginia took the exchange offline for more than two hours, capping a brutal week that already included a 14% workforce cut.

Coinbase Posts US$394M Q1 Loss As AWS Cooling Failure Knocks Exchange Offline The Same Day
Image via pymnts.com

Key Takeaways

  • 1.The Nasdaq-listed exchange reported a US$394 million GAAP net loss on revenue of US$1.41 billion, down 31% year-on-year, in numbers released after the bell on Wednesday.
  • 2."We're leading on the next frontier with over 90% of on-chain agentic stablecoin transaction volume happening on Base," he said.
  • 3.Subscription and services revenue, the line Coinbase has been pushing as the future of the business, came in at US$584 million, or 44% of the total.

Coinbase used its Q1 2026 earnings to talk about reinvention. The market replied by treating the report as one more reason to mark the stock lower.

The Nasdaq-listed exchange reported a US$394 million GAAP net loss on revenue of US$1.41 billion, down 31% year-on-year, in numbers released after the bell on Wednesday. Subscription and services revenue, the line Coinbase has been pushing as the future of the business, came in at US$584 million, or 44% of the total. Stablecoin revenue alone contributed US$305 million.

CEO Brian Armstrong framed the result as a quarter where the underlying franchise kept growing even as the price of the assets it sells fell sharply.

"We executed well on what was in our control in Q1," Armstrong said. "We hit a new all-time high in USDC held in Coinbase products and saw 10x year-over-year growth in stablecoin transactions on Base."

Chief Financial Officer Alesia Haas pinned the headline numbers on the broader crypto selloff rather than anything specific to Coinbase. "The market environment this quarter was softer, but the underlying fundamentals of our business remain strong," she told investors.

Armstrong went further on AI, signalling that he sees Coinbase's future not as a retail trading venue but as the rails for autonomous on-chain commerce. "We're leading on the next frontier with over 90% of on-chain agentic stablecoin transaction volume happening on Base," he said.

That strategic pitch arrived less than 48 hours after Coinbase informed roughly 700 staff, about 14% of its workforce, that they would be cut. Armstrong's internal memo described the goal as becoming "lean, fast and AI-native", with manager layers replaced by what he called player-coaches.

The day did not get any kinder once trading opened. On Thursday morning Asia time, Amazon Web Services confirmed a thermal event in its US-EAST-1 region in Northern Virginia, in a single availability zone known as use1-az4. The cooling failure caused a power loss that damaged hardware and forced AWS to throttle services in the affected zone.

Coinbase, which runs significant infrastructure on AWS, was one of the highest-profile victims. Core exchange functions, including order placement and account access, ran in degraded mode for more than two hours, with some users reporting full lockouts. The company emphasised customer funds were not at risk and pinned the disruption squarely on AWS. CME Group's electronic trading platform also reported issues during the same window before restoring normal access.

The combined picture, a revenue collapse, a strategic pivot announced via mass layoffs, and a multi-hour outage on a key infrastructure provider, was enough to drag the Coinbase stock lower in extended trading and reignite an old debate about whether the company is too dependent on a handful of cloud regions.

For Armstrong, the bet is clear. Stablecoins, USDC and Base now sit at the centre of the Coinbase narrative, not Bitcoin trading volumes. The challenge is that the AI-native, stablecoin-first business he is describing still has to run on someone else's data centres, and one of those data centres just spent a morning struggling to keep itself cool.