Alibaba: State-Led Data Center Could Reduce Cloud Revenue

We’ve reduced our fair value estimate of Alibaba stock.

The Alibaba logo and signage is displayed on a building in Xixi, Hangzhou, China.
Alibaba Group via Alibaba Group

Key Morningstar Metrics for Alibaba

  • Fair Value Estimate
    : USD 241.00
  • Morningstar Rating
    : ★★★★★
  • Morningstar Economic Moat Rating
    : Wide
  • Morningstar Uncertainty Rating
    : High

Alibaba BABA shares fell 27% over the last month on three developments: an unconfirmed report of a CNY 2 trillion state-led data center buildout in China, US legislative threats penalizing Chinese AI model distillation, and a Pentagon blacklist designating it a “Chinese military company.”

Why it matters: Such a state-led data center buildout is likely to target government and state-enterprise artificial intelligence workloads. This could reduce Alibaba’s public-sector compute demand, prompting a 10% annual cut to our cloud revenue forecasts starting in fiscal 2030, the assumed completion year.

  • Following trade war precedent, we think Beijing will retaliate—potentially banning critical exports—if US laws hinder China’s AI model advancement through penalizing AI distillation. Hence, we don’t expect a long-term negative impact on the advancement of China’s AI model capabilities.
  • A new law bans the US Defense Department from working with firms that employ lobbyists representing Pentagon-blacklisted “Chinese military companies.” The Japan Times reported that Alibaba lost five lobbying firms. We see limited impact, as Alibaba has a negligible US footprint.

The bottom line: We cut wide-moat Alibaba’s fair value estimate by 7% to USD 241 per ADS (HKD 235 per share) due to a reduction in cloud revenue. Shares are cheap at a price/fair value estimate of 0.6, due to longer-standing concerns over on-demand delivery losses and surging AI compute costs, both of which we think are overdone.

  • We believe a strong position in on-demand delivery is necessary for Alibaba to maintain its online retail market position, and that the business will be profitable by fiscal 2029.
  • While the cost and the amount of compute for training models have surged, we think this is at least partially offset by improving model architecture and algorithms, and the adoption of more cost-efficient domestic and proprietary T-Head chips.

Editor's Note: This analysis was originally published as a stock note by Morningstar Equity Research.

The author or authors do not own shares in any securities mentioned in this article. Find out about Morningstar's editorial policies.