Bernstein: CoreWeave (CRWV.US) may be most affected if AI model development slows down

According to Golden Ten Data, investment bank Bernstein stated that the slowdown in artificial intelligence (AI) model training called for by industry giants could have a negative impact on data center developers and new cloud service providers, with CoreWeave (CRWV.US) likely to be one of the most affected companies.

Anthropic CEO Dario Amodei called for a slowdown in the development of cutting-edge AI models in a cautionary article published on September 12. He stated bluntly that the risks brought by AI are “severe” and that time must be devoted to addressing these risks. The appeal was quickly echoed by two key figures. Musk reposted Amodei’s post on social media platform X, commenting: “Dario is right.” OpenAI’s leader Altman also wrote on X: “I agree with Dario, we need to control the pace of frontier AI development.”

Bernstein analyst Maidson Rezaei said in an investor report on Monday that a slowdown in model training speed or quantity would mainly affect data centers in rural areas. The analyst said: “From the perspective of infrastructure and new cloud service providers, a decline or slowdown in training demand will result in demand draining away from rural data center locations, many of which are purpose-built for these latency-insensitive training needs.”

The analyst added: “In the US, we have a total nameplate capacity of 488 GW in data center development project reserves, of which 170 GW is deemed credible. Of this 488 GW reserve, 36% is located in rural areas and another 34% in Tier 3 markets (such as West Texas). This means that 70% of the development project reserves are built around training demand (or latency-insensitive inference demand).”

Based on this analysis, CoreWeave has the most significant risk exposure. The analyst stated: “We estimate that 25% of CoreWeave’s currently operational power capacity in the US is located in Tier 3 and Tier 4 markets. In addition, about 74% of its contracted power capacity is also in these markets. Its backlog mainly consists of ‘take-or-pay’ contracts, so we expect this portion to remain unaffected; however, if the pace of training development slows, we may see a decrease in demand for the contracted but unsold rural power capacity.”

In contrast, data centers located in metropolitan and smaller urban areas appear to be best positioned to withstand the impact of a slowdown in training demand. This includes Equinix (EQIX.US), Digital Realty Trust (DLR.US), and Csquare (CSQR.US). These companies have 95%, 92%, and 94% of their data center capacity located in large or small urban regions, respectively.

Bernstein has given CoreWeave an “underperform” rating with a target price of $74. Meanwhile, the investment institution has rated Csquare, Digital Realty, and Equinix as “outperform,” with target prices of $27, $226, and $1,270 respectively.

Bernstein pointed out that even if data centers shift from a training-oriented to an inference-oriented approach—a transformation necessary for achieving more agentic workflows—data centers based in urban areas “remain the safest and most valuable.”