Stay updated with the latest stock market news and insights
Weekly insights + SMS alerts
One of the most revealing comments from Alibaba Group Holding Ltd. (BABA)'s fiscal first quarter earnings call wasn't about revenue or artificial intelligence demand.
Instead, it was about the staying power of Nvidia Corp. (NVDA)'s older AI chips, with management saying GPUs purchased as far back as 2018 are still operating at full capacity across its AI infrastructure.
Discussing the economics of the company's AI investments, Chief Financial Officer Toby Xu said Alibaba expects its AI assets to generate "very positive and robust cash flow" after a three-year payback period.
To illustrate the point, Xu cited the company's existing GPU fleet, saying "an A100 purchased in 2020 or a V100 purchased in 2018 even today are still running at full capacity."
The comment offers a rare glimpse into the useful life of AI accelerators inside one of the world's largest cloud providers. While much of the industry's attention has centered on the rapid rollout of newer chips, Alibaba indicated that older hardware continues to play a meaningful role in serving AI workloads.
Rapid GPU turnover has fueled fears that today's cutting-edge accelerators could become obsolete within just a few years.
Alibaba's experience suggests otherwise. Rather than retiring older GPUs as newer chips arrive, the company says its existing hardware remains fully utilized years after deployment — and, per Xu, continues to generate cash flow well past its roughly three-year payback period.
Alibaba didn't disclose what share of its AI workload still runs on V100s or A100s. But continued full utilization of both generations suggests demand has been strong enough to absorb legacy and new hardware alike.
Alibaba's remarks may carry implications beyond its own cloud business. If sustained utilization of older accelerators holds up across other large-scale deployments, it would ease concerns that rapid chip advances are quickly eroding the value of existing GPU fleets.
Investors tracking Nvidia and the broader AI infrastructure trade should watch upcoming hyperscaler earnings for similar disclosures. Confirmation from other cloud providers would reinforce the case that AI demand is strong enough to extend hardware's economic life rather than render it obsolete overnight.
Real-time alerts on price moves, news, and trading opportunities.
Join 20,000+ investors. No spam, ever.
