The news of the (extremely) cheaper Chinese DeepSeek response to OpenAI's ChatGPT—allegedly "40-50x more efficient than other large language models," according to Goldman's Rich Privorotsky—has triggered a global selloff across the entire AI ecosystem.

The selloff spans chip designers, AI firms, semiconductor equipment manufacturers, AI data center companies, and "Powering Up" baskets tracked by Goldman. "If you can do more with less, it naturally raises the question of whether so much capacity is necessary," Privorotsky noted.

Here are Privorotsky's comments on DeepSeek: 

Markets sharply lower driven by tech. DeepSeek is something we are going to be talking about a lot this week and especially ahead of Mag 7 earnings where the focus will be on upcoming AI related capex outlays. What makes DeepSeek important is not necessarily that it's Chinese or rapidly ascended to one of the most popular downloads in the Apple App Store but that it appears quite a fundamental advancement in the efficiency of AI technology. For a lot of complicated reasons it's basically a huge upgrade to effectiveness of inference. By some measurements, 40-50x more efficient than other models. If you can do more with less it naturally leads the question to whether you need so much capacity. Obviously this will be a huge area of debate and I am far from an expert the nuances of semiconductor equipment or AI data layer inference...but given the immense amount of market cap added on this thematic I think we hit hard in reverse in the tends of last week. The power up baskets are going power down. Semis/ai/tech are going to be under pressure today. Think maybe now we understand why NVDA has been de-rating to the extent it has.

The whole "do more with less" theme sparked by DeepSeek will likely suggest that upcoming AI-related capex outlays will be greatly paired down and may prompt revisions to peak data center capacity forecasts produced by various institutional desks. Most notably, we'll look at the forecast formulated earlier this month by Goldman's James Schneider, Michael Smith, and others: 

Investor interest in the datacenter market has increased dramatically, particularly with respect to the demand and supply dynamics of this rapidly evolving market. On the demand side, the explosion of activity tied to AI has led to a spending "arms race" where hyperscalers and corporates alike are building LLMs which need to be trained on vast quantities of information using GPUs which require unprecedented amounts of power. On the supply side, hyperscalers, datacenter operators, and asset managers are deploying unprecedented amounts of capital to build new high-capacity datacenters against a backdrop of scarce power supply provided by the traditional grid.

In this report, we introduce a model that quantifies the global datacenter supply and demand picture over the next five years based on the pipeline of planned datacenters we see on a regional basis. We also forecast the demand we see from different types of workloads: AI, hyperscale/cloud, and traditional corporate. By matching these supply and demand dynamics, we forecast datacenter occupancy and utilization rates over the medium term.

Our conclusion: Our base case forecast estimates 93% occupancy as of 1Q25. In this base case scenario, we estimate that the data center supply/demand balance will tighten further through 2H2026 - followed by gradual loosening beginning in 2027 as more supply comes online.

Privorotsky told clients: "From a few conversations I have had with various investors this morning, the initial thoughts are similar - negative capex impact and a meaningful change of narrative in the US AI space is worrisome, in terms of R and I in ROI." 

Returning to the data center capacity forecast, with peak demand previously projected for late 2026, those estimates may need reevaluation.

 Global data center occupant rate... 

This likely explains why the entire AI ecosystem has taken a significant leg down today—from high-end chip companies and data centers to AI stocks, AI infrastructure-related firms, and the 'Powering Up' basket tracked by Goldman. 

Latest DeepSeek reporting:

And we'll leave you with this...

. . .