Nebius just weakened a big neocloud bear case, that older GPUs quickly become economically obsolete. From October 1 it raises on-demand prices about 17-21% across its NVIDIA fleet, with H100 rates up from $2.95 in May to $4.50, about 53%. It is also signing longer contracts at about $20M-$25M per MW with prepayments and raising AMD CPU and memory prices. Compute is still tight, but on-demand pricing is cyclical and Blackwell and Rubin add supply. Next, watch for pricing in ARR. x.com 

Nebius can raise GPU rents despite cheaper tokens because lower task costs spur far more usage and compute is a small part of Anthropic's revenue. Its S-1 puts compute at ~15 cents per Claude dollar; a hypothetical 20% rent rise would cut first-half 2026 gross margin from 85% to ~82%, up from 40% in 2025. Opus 5.5 cut cost per task ~40%, yet customers spending $1 a year ago spend ~$9 now; H200 rents rose ~160% this year. Nebius raised prices 17-21% on Oct 1 and has reservations into 2028. x.com 

Nebius's integrated AI cloud offers a substantial long-term opportunity beyond simply controlling powered land as training and inference demand expands. It builds data centers and GPU clusters and runs networking, storage and customer software; Microsoft and Meta have multi-year commitments. The correction after a 10x rise is unfolding, but deeper price targets have not been reached; the next 6-12 months could offer a more attractive entry than near the highs. x.com 

Nebius's price increases show the GPU compute deflation thesis is empirically dead. It just raised H100, H200, B200 and B300 pricing 17-21%, its second hike in four months, leaving those GPUs up 53-56% since June. Owners of uncontracted capacity don't raise prices twice in a cycle when supply is loose; compute deflation was a model assumption, not a market observation. x.com 

Nebius raising prices on compute is bullish. Per @AIStockSavvy, Nebius will raise on-demand Nvidia GPU rates from October 2026, and the stock rose 2.5% after hours on the news. x.com  x.com 

Nebius' business model faces pressure once GPU supply and demand normalize, perhaps within five years. It faces enormous supplier power, more competitors and likely rising buyer power, and lenders and equity investors may view it differently once utilization and pricing stop being exceptional. China could flood global markets with low-cost compute, and the market already seems to price a very long AI boom. The case weakens if AI demand proves durable, like an 18-year real estate cycle. x.com 

AI agents' productivity over human labor should lift GPU demand and rates for neoclouds. The bottom looks in for CoreWeave, Nebius, Boost Run, Sharon AI and IREN. Gavin Baker says token prices can fall as revenue per GPU rises; paraphrasing Elon, even $38 per GPU-hour can pay off. CoreWeave's five-year fixed contracts may lag current rates; Boost Run has less capacity locked in. If SpaceX takes more growth, slower cash burn could rerate neoclouds, though smaller customers may pay less. x.com  … Show more

Nebius's bet on short-term GPU rentals is working, letting it raise on-demand rates from October 1, per the quoted announcement. H100 pricing rises from $3.85 to $4.50 per GPU-hour, H200 from $4.50 to $5.40, B200 from $7.15 to $8.50 and B300 from $7.85 to $9.50. x.com 

CoreWeave and Nebius would reprice first whenever AI labs stop paying up for certainty, because the compute premium sits on energized contiguous capacity, not GPUs. Per a Google CTO on an AlphaSense expert call, SpaceX charges $30-50M per MW, 2-3x other neoclouds, for energized gigawatt blocks. CoreWeave and Nebius are writing shorter tenors, and no lab wants to be locked into 2026 clearing prices past 2028. Hyperscalers with their own silicon have the most cushion. x.com  x.com 

Nebius can make vastly more money when Inferize cuts wasted GPU time and spot prices rise above the recently removed caps. Together, these changes unlock what looks like insane value. x.com