The "land scramble" for data center capacity continues, but after supply and demand reach equilibrium, Neocloud may be the first to come under pressure.

The "land scramble" for data center capacity continues, but after supply and demand reach equilibrium, Neocloud may be the first to come under pressure.

Data center capacity is rapidly being locked in units of “gigawatts,” and what buyers are competing for is not the concept of the data room, but available power, delivery speed, and controllability. During the supply crunch window, short-cycle solutions like Neocloud (emerging cloud service providers) can still secure large orders, but once supply eases around 2028, they may be the first to feel the pressure.

The latest example comes from CoreWeave. According to Chase Wind Trading Desk, Bernstein noted in a recent research report that CoreWeave disclosed two new contracts last week: a $21 billion incremental contract with Meta and a multi-year agreement with Anthropic (amount undisclosed). These signings exceeded its expectations for early 2026.

The order news continues the market’s pricing logic of “computing power as asset,” but Bernstein has not turned optimistic as a result. The firm maintains an “Underperform” rating for CoreWeave. It believes that as the supply-demand gap shrinks, Neocloud will be the most impacted segment because it is easier to exit.

It is estimated that global data center supply is currently about 100GW, with the market expected to reach around 190GW in the next five years, and possibly about 220GW by 2030 under a scenario of buoyant AI demand. In the short term, the “land grab” continues, but this competition does not mean all GW are equally scarce and equally secure.

CoreWeave signs new deals again, Neocloud remains a “patch” during shortage cycles

The $21 billion incremental contract between CoreWeave and Meta, together with the new agreement with Anthropic, are not isolated incidents but typical behavior under tight capacity: almost every day someone is signing, building, or financing new data center capacity to meet demand.

For investors, the short-term logic is relatively clear: when self-building and long-term leases have lengthy delivery cycles, the ability to quickly provide GPU cloud resources still gives bargaining power, which also explains why the market may continue to respond positively to “deal signing news” in the coming months.

But this reaction is more cyclical than proof of structural moats.

Why Hyperscalers are still buying Neocloud: speed, optionality, and sandboxes

A set of explanations for why Hyperscaler (large-scale cloud service providers) use Neocloud centers on “speed” and “optionality.”

First is existing power and delivery speed. CoreWeave has “legacy power from the mining era,” and swallows incremental capacity quickly by leasing data centers for very long terms, thus holding an advantage in delivery speed.

Second is lower short-term commitment. Compared to self-building or signing a 15-year lease, a 5-year Neocloud contract is more like “low-commitment bandwidth,” suitable for gaining a computing power window when AI demand visibility is still unclear.

Third is sandbox and learning opportunity. CoreWeave, as “the largest pure AI play,” is one of the most mature GPU cloud operators, so Hyperscalers may use this for experiments and accumulating experience before deciding whether to self-build or deploy at large scale.

“Gigawatts” are also layered: four levels from self-build to Neocloud

Buyers will rank different models based on workload types and financial constraints. In the current tight environment, the four modes occur in parallel, but the ranking determines risk exposure after a cycle turning point.

Their priorities are:

1. Self-build and hold real estate in suitable regions as much as possible, to match long-term workloads and inference scenarios.

2. Whole-building leasing (powered shell), balancing capex and opex, and striving for faster launches.

3. Small-scale wholesale colocation, meeting specific market demands without taking on entire buildings.

4. Using traditional cloud or Neocloud as a “band-aid” when capacity can’t be obtained in time, the location happens to match, or only short-term experiments are needed.

As supply starts to ease (but “not yet now”), the fourth category will be “fastest and most impacted,” because exiting Neocloud is easier than exiting a data center—contracts are shorter and there’s no need to relocate hardware.

The supply turning point points to 2028, structural pressure for Neocloud may break out later

It is expected that the tightness in data center capacity will ease in 2028, and may be delayed to 2029 if AI demand is extremely optimistic. The long-term view is that Hyperscalers will ultimately form a more “hybrid” capacity strategy, but when supply is no longer extremely scarce, Neocloud will find it harder to continuously sign large-scale hyperscaler-level contracts.

This also constitutes the core difference regarding CoreWeave: maintain “Underperform,” and state that its long-term revenue expectations are significantly below market consensus, and worry that after 2028, as supply becomes more abundant, Hyperscalers will prefer self-building and long-term leases, possibly also competing directly with Neocloud in enterprise business.

Meanwhile, CoreWeave’s claim of a software moat withstanding challenges from Hyperscalers’ scale and investment is questionable.

However, Bernstein also acknowledges that the timing may be postponed. The firm expects CoreWeave to still add about $45 billion in new contracts before the end of 2027. The tide will not shift quickly, meaning the market may continue to trade for quite a long period around "how much GW each secures."

 

~~~~~~~~~~~~~~~~~~~~~~~~

The above exciting content comes from Chase Wind Trading Desk.

For more detailed interpretations, including real-time analysis and frontline research, please join【Chase Wind Trading Desk ▪ Annual Membership】

Risk Warning and DisclaimerThe market carries risks, and investments must be made cautiously. This article does not constitute personal investment advice, nor does it take into account any individual user's special investment objectives, financial condition, or needs. Users should consider whether any opinions, views, or conclusions herein are suitable for their specific circumstances. Investment based on this is at your own risk. ```