The Neocloud Trojan Horse

One of the most frequent pushbacks that I have received to my recent posts on AWS (see here, and here) is that the neoclouds are not a legitimate concern for hyperscalers such as AWS and Azure in the long term. Frankly speaking, I myself operated under this assumption for the last couple of years. What made me eventually question such comforting hypothesis is the capex outlook of CoreWeave and Nebius for 2026. For context, CoreWeave guided $30-35 Billion capex whereas Nebius guided for $16-20 Billion capex for 2026. I don’t know about you, but I would be at least somewhat concerned about long-term industry structure of cloud if someone mentioned to me that neoclouds capex would be almost one-third of AWS capex size by 2026.

It also doesn’t help that almost on a weekly basis we are seeing Nvidia providing more funding to these neoclouds or/and these neoclouds are partnering with major tech companies to become more relevant players in the AI workloads. For example, just today Nebius announced that Meta agreed to buy $12 Billion of AI capacity by 2027 and an option on another $15 Billion over five years. At this point, it seems almost foolhardy to consider these neoclouds fringe or irrelevant players.

In fact, if you listen to these neoclouds earnings calls or recent Morgan Stanley TMT conference sessions, they seem to be not oblivious about the bear cases against their business model. I will share some key takeaways from these calls behind the paywall.

In addition to “Daily Dose” (yes, DAILY) like this, MBI Deep Dives publishes one Deep Dive on a publicly listed company every month. You can find all the 66 Deep Dives here.