SemiAnalysis 称英伟达限制采用 AMD 与 TPU 的 neocloud 参加 GTC Berlin
Nvidia reportedly refused some neoclouds access to GTC Berlin for adding AMD and TPU chips
SemiAnalysis 称英伟达拒绝部分采用 AMD GPU、TPU 等第三方芯片的 neocloud 厂商参加 GTC Berlin，以此施压其减少采用竞品硬件。报道提到超大规模云厂商因采购量更大未受同样对待，部分 neocloud 已低调部署 AMD 方案以规避压力。SemiAnalysis 今年早些时候还称英伟达会推迟向这类厂商分配新芯片，并影响其 IPO 与融资支持；也有观察者认为英伟达有权决定自家会议的邀请名单。
推荐理由
AI 评分 62
SemiAnalysis 的说法把芯片供应之外的限制手段摆上台面，读者可据此观察云厂商与 GPU 厂商的议价关系。
主题
正文
当前显示原文，中文全文译文尚未就绪。
Rumor mill: Analysts with SemiAnalysis have uncovered what could be another troubling trend in the AI hardware industry. Nvidia is reportedly retaliating against some "neocloud" providers for integrating GPUs and other chips from rival manufacturers into their cloud offerings.
SemiAnalysis' latest "shocking" rumor allegedly confirms what the firm had already revealed in June. Several neocloud providers have reportedly been denied access to GTC Berlin, Nvidia's upcoming conference covering AI, GPUs, and related technologies. The Jensen Huang-led company apparently disapproves of cloud providers adding AMD GPUs, TPUs, and other third-party silicon to their offerings, and is making it harder for them to attend one of the industry's most important events.
Hyperscalers have apparently been spared Nvidia's "punishment" because they wield considerably more purchasing power than neocloud providers. The term neocloud is gaining traction as a way to describe a specific category of modern cloud providers that offer specialized IT infrastructure designed to efficiently handle AI workloads.
According to SemiAnalysis, neocloud providers are feeling Nvidia's pressure as the GPU maker attempts to discourage them from adopting competing GPU and AI hardware. The situation is reportedly so tense that some neocloud providers are now deploying AMD-based solutions without publicizing the move. This "quiet" approach might help them avoid further retaliation from Nvidia.
– SemiAnalysis (@SemiAnalysis_) October 16, 2026
SemiAnalysis first disclosed Nvidia's alleged tactics earlier this year. At the time, the "silicon intelligence" company stated that Nvidia was punishing neocloud providers by withholding early allocations of upcoming chips. Financial support for IPOs and capital-raising initiatives was also reportedly at risk of drying up. Nvidia has directly funded some neocloud providers, which is why some observers have dismissed SemiAnalysis' latest claims as a non-story.
Nvidia launched its GTC conference in 2009, bringing together engineers, researchers, and developers for an event that has since become heavily focused on AI. Some observers also argue that the Santa Clara-based company is entitled to decide whom it invites to its own conference.
Nvidia has repeatedly faced accusations of abusing its dominant position in the GPU market to disadvantage competitors. In 2024, a former AMD GPU chief accused the company of behaving like a "GPU cartel" by delaying shipments to data centers, OEMs, and even resellers.
The AI boom is turning GPUs and other specialized accelerators into highly sought-after hardware. AMD is also benefiting from the trend, particularly in the data center market, where GPU manufacturers continue to struggle to meet demand from hyperscalers and other cloud providers.