Amazon said most of its cloud services were back online by Friday after overheating at one of its data centers caused an outage that affected customers, including Coinbase.

The company said a sudden rise in temperatures at a single Northern Virginia facility led to a power disruption. The company added that full recovery would still take several hours.

According to Coinbase, its services had been fully restored after the incident temporarily affected availability.

Overheating in data centers is a major problem for companies as advanced AI and cloud servers consume massive amounts of power and generate intense heat. To regulate the heat, data center operators are increasingly turning to water or specialized coolants, which are thousands of times more efficient than traditional air cooling.

Thursday’s incident was the second major overheating-related disruption in recent months. In November last year, CME Group CME.O experienced one of its longest outages in years after a cooling failure at data centers operated by CyrusOne.

At 8:12 a.m. ET, outage reports for AWS on Downdetector, an outage tracking website, had dropped to 72, down from nearly 600 on Thursday night.

AWS has ‌been ⁠bringing additional cooling system capacity online but said it was taking longer than expected to add the capacity required to safely restore all remaining affected systems.

The cloud computing platform also said it had shifted traffic away from the impacted Availability Zone for most services. An "Availability Zone" comprises one or more connected physical data centers and are designed to operate ⁠independently within an AWS Region.