Nvidia has reportedly denied access to its GTC Berlin 2026 conference for several smaller cloud providers, known as neoclouds. According to a report by SemiAnalysis published on October 16, 2026, the decision stems from these companies incorporating AMD GPUs and Google TPUs into their infrastructure.
The GTC Berlin event is scheduled for October 20–22, 2026. Nvidia is allegedly pressuring these providers to maintain exclusive Nvidia-based fleets. This policy appears selective, as major hyperscalers like AWS and Microsoft Azure, which possess greater purchasing power, are not facing similar restrictions.
As of August 2026, many neoclouds were splitting their hardware orders between Nvidia and AMD on a near 1:1 basis, with at least two providers deploying 10 000 units from each vendor. While Nvidia retains a performance advantage, providers are increasingly turning to AMD for cost efficiency.
The neocloud sector has expanded significantly, reaching over 300 providers by September 2026, a 55% increase from the previous year. Industry revenue exceeded $25 billion in 2025, with projections estimating the market could reach $400 billion by 2031.
Nvidia is expected to capture 75% of the AI accelerator market in 2026, a decline from 87% in 2024. In response to the current pressure, some neocloud executives are considering switching suppliers or offering non-Nvidia chips without public disclosure to avoid conflict.
