U.S. Stocks Dip Amid Concerns Over AI Demand
U.S. stocks related to artificial intelligence experienced a downturn recently, with Microsoft’s shifting strategies in the data center market raising questions about the trajectory of AI demand. The Philadelphia Semiconductor Index saw a decline, and several major tech companies were affected.
According to a report from TD Cowen, a research firm, Microsoft has reportedly scaled back agreements with data center operators in the U.S. for substantial power capacity. This pullback has extended into European markets as well, signaling a potential reassessment of the anticipated growth in AI-related infrastructure.
The TD Cowen report noted that the changes by Microsoft might be connected to OpenAI, the creator of ChatGPT, in which Microsoft holds a stake. Microsoft’s data center in Wisconsin was intended initially to support OpenAI’s operations, a fact previously confirmed by TD Cowen. However, the emergence of DeepSeek this year has led the industry to rethink the AI sector, possibly influencing the company to reassess data center needs.
Key Players Show Declines
On the day of the report, the Philadelphia Semiconductor Index closed down by approximately 3.3%. The impact was felt by industry leaders, with Nvidia falling more than 5.7%, leading the declines among the seven major tech companies. AMD’s shares also decreased, dropping by almost 8.9%, illustrating the broad impact of the market’s reaction to the developments.
TD Cowen’s channel survey indicated that Microsoft has postponed or cancelled data center leases both in the U.S. and Europe. These actions, occurring over the past six months, encompass over 2 gigawatts of capacity undergoing leasing processes in the U.S. and Europe. Despite these adjustments, Microsoft has reiterated its commitment to maintaining its $80 billion investment plan.
Industry Concerns
Earlier in the week, Alibaba Chairman Joe Tsai warned about a potential bubble in data center construction. He suggested some projects might be outpacing the demand for AI services. Tsai noted that there seems to be significant overlap in data center investments, especially in the U.S., resulting in resource inefficiency and increased market competition. This sentiment is reinforced by the fact that some entities are still seeking capital through SPACs to build data centers, despite not having secured user agreements.
Additionally, a Goldman Sachs analyst team has revised downward the projections for the computing power devoted to training AI and the shipments of AI training servers. Specifically, the shipments for 2025 and 2026 have been adjusted from 31,000 units and 66,000 units down to 19,000 units and 57,000 units, respectively.
Trends in Data Center Industry
OpenAI is increasingly looking to acquire data center capacity from third-party providers, including GPU-as-a-Service suppliers and third-party data center operators, which include CoreWeave. Furthermore, OpenAI may develop its own data centers long-term, looking to establish multiple “Stargate” projects with significant capacity requirements.
Microsoft’s Advances in Cooling Technology
In October of last year, Microsoft announced plans for data center operators to retrofit existing facilities to support liquid cooling technology. Recent surveys indicate that Microsoft is developing an internal plan aimed at improving the rack density of traditional cloud data centers by implementing air-assisted liquid cooling solutions. This approach is considered more cost-effective than a full-scale retrofit, allowing for targeted cooling of high-density workloads.
