Nvidia's flagship conference season has taken an awkward turn. A report published by SemiAnalysis on October 16, 2026, states that several neoclouds were denied access to Nvidia's GTC Berlin 2026 event after adding AMD GPUs and Google TPUs to their fleets.

The timing is notable. GTC Berlin is scheduled for October 20 to 22, meaning the affected providers learned of their exclusion only days before the doors open.

What the SemiAnalysis report says

Neoclouds are smaller cloud companies built largely around renting out GPU computing power, primarily for AI workloads. According to SemiAnalysis, Nvidia's reported pressure reflects an effort to keep these smaller providers on Nvidia-only fleets—devoid of AMD Instinct accelerators and Google TPUs, the custom machine learning chips Google designed in-house.

The pressure appears selective. Per the report, Nvidia has not employed the same tactics against hyperscalers such as AWS and Microsoft Azure, which command far greater purchasing power. Those giants also have less need to rely on Nvidia alone: both AWS and Microsoft, like Google, design their own custom AI chips in-house.

The neoclouds, however, are not simply complying. SemiAnalysis reports that some providers are weighing a switch of suppliers to escape the pressure, while certain neocloud executives are considering quietly offering non-Nvidia chips in an effort to limit potential backlash from Nvidia.

Why neoclouds are diversifying

The diversification predates the reported exclusions. As of August 2026, many neoclouds were reportedly splitting their orders between AMD Instinct and Nvidia GPUs on a near 1:1 basis, and at least two providers were deploying 10,000 GPUs from each vendor.

The reasoning is largely economic: AMD holds a price advantage, while Nvidia retains the performance lead. Competition within the sector has also intensified. As of September 2026, more than 300 GPU providers operated in the neocloud space, up 55% from roughly 209 a year earlier.

The money behind the fight

Neocloud revenues exceeded $25 billion in 2025, with projections placing the sector near $400 billion by 2031. That projected growth underscores how much commercial weight rests on the chip-sourcing decisions at the center of this dispute.

Nvidia's overall position remains dominant, though its lead is narrowing. The company is projected to capture around 75% of AI accelerator revenue in 2026, down from 87% in 2024.

Background: GTC as more than a conference

GTC—short for GPU Technology Conference—is Nvidia's flagship developer and industry event series, traditionally the stage for the company's most significant hardware and software announcements. The Berlin edition is set to run across venues including the Tempodrom and STATION-Berlin.

What this means for Nvidia, AMD and the neoclouds

The report already points to some providers considering a supplier switch in response to Nvidia's pressure. The quieter approach some executives are reportedly weighing—offering non-Nvidia chips without fanfare—suggests many would prefer to avoid open conflict while still hedging.

For those tracking AI infrastructure, three concrete signals stand out: whether Nvidia's projected 75% share of AI accelerator revenue in 2026 holds or slips further; whether more neoclouds publicly disclose their AMD or TPU deployments, or go quiet about them; and whether Nvidia extends, softens or abandons the reported pressure once GTC Berlin wraps on October 22.