Neoclouds target AI workloads but struggle with market clarity
Neoclouds are specialized cloud providers built around GPU compute, high-bandwidth memory, and fast interconnects for AI training and inference. Their revenue is growing rapidly, with forecasts pointing to hundreds of billions of dollars by the early 2030s. Despite the growth, the article says enterprises often overlook neoclouds when redesigning AI infrastructure, partly because the category's role remains unclear.
Neoclouds are specialized AI clouds offering GPU compute, high-bandwidth memory, and fast interconnects. They deploy dense, liquid-cooled clusters with high-speed networking, often coming online in months rather than the three-to-five years typical for hyperscale builds. Revenue reached $9 billion in Q4 2025, up 223% year over year, and exceeded $25 billion for the year. Forecasts suggest nearly $400 billion by 2031 at 58% compound annual growth.
ABI Research expects GPU-as-a-service revenue from these providers to exceed $250 billion by 2030, up from $42 billion in 2025, with over 2,200 neocloud data centers by 2035 versus 558 in 2025. CoreWeave surpassed $5 billion in annual revenue faster than any cloud platform. Much growth came from large tech-to-tech deals; Microsoft accounted for 62% of CoreWeave's 2024 revenue, while Nvidia is a key Lambda and CoreWeave customer.
Count? First para: Neoclouds(1) are2 specialized3 AI4 clouds5 offering6 GPU7 compute8 high-bandwidth9 memory10 and11 fast12 interconnects13. They14 deploy15 dense16 liquid-cooled17 clusters18 with19 high-speed20 networking21 often22 coming23 online24 in25 months26 rather27 than28 the29 three-to-five30 years31 typical32 for33 hyperscale34 builds35. Revenue36 reached37 $9 38 billion39 in40 Q4 41 2025 42 up43 223%44 year45 over46 year47 and48 exceeded49 $25 50 billion51 for52 the53 year54. Forecasts55 suggest56 nearly57 $400 58 billion59 by60 2031 61 at62 58%63 compound64 annual65 growth66. Good 66.
Second: A
If neoclouds clarify their role, enterprises may gain more tailored AI infrastructure options, potentially speeding model training and inference while reducing reliance on a few hyperscalers. Workers and communities near new data centers could see economic activity, but also pressure on power and water. Buyers may benefit from competition and lower costs, though unclear positioning could keep procurement concentrated among established providers. The outcome depends on whether neoclouds communicate distinct value and whether their rapid buildout addresses environmental and reliability concerns. Count: If1 neoclouds2 clarify3 their4 role5 enterprises6 may7 gain8 more9 tailored10 AI11 infrastructure12 options13 potentially14 speeding15 model16 training17 and18 inference19 while20 reducing21 reliance22 on23 a24 few25 hyperscalers26. Workers27 and28 communities29 near30 new31 data32 centers33 could34 see35 economic36 activity37 but38 also39 pressure40 on41 power42 and43 water44. Buyers45 m