The real story of GTC 2026 is not Nvidia’s next chip. It is how far behind the rest of Big Tech is on AI infrastructure. When Jensen Huang takes the stage, the subtext is that Amazon, Google, and Microsoft still depend on a single vendor for the bulk of their AI compute, and that dependency is the story.
GTC 2026 will reveal how far behind the rest of Big Tech is on AI infrastructure
GTC is Nvidia’s flagship annual event, and as TechCrunch has reported, the 2026 keynote is where Huang typically announces new products, partnerships, and the company’s vision for the future of computing and AI. But the conference also functions as a mirror: it shows who has kept pace with Nvidia and who has not. Microsoft, Amazon, Google, and Meta are Nvidia’s largest customers. According to 24/7 Wall St and Yahoo Finance, Microsoft alone accounts for roughly 19% of Nvidia’s annual revenue and spends 47% of its capital expenditures on Nvidia chips. Collectively, Big Tech is spending hundreds of billions on AI infrastructure, and a large share of that flows through Nvidia. The AI infrastructure market has been estimated at over $200 billion, with roughly 70% flowing through just a handful of companies: Nvidia for chips, and Microsoft, Amazon, and Google for cloud. So when GTC 2026 rolls around, the real takeaway is how dependent the biggest players in tech have become on one company for the critical layer of the AI stack. The next chip is a headline; the dependency is the structural fact.
That dependency has not gone unchallenged. Analysts and reports from Wedbush, Yahoo Finance, and The Meridiem describe a “Great Silicon Divorce” and a shift toward “Silicon Sovereignty”: Amazon, Google, and Microsoft are building their own AI chips to reduce reliance on Nvidia and cut costs by an estimated 30% to 40%. Google’s TPU v7 now handles over 75% of Gemini computations; Amazon has launched Trainium3; Microsoft is ramping production of its Maia 200. Meta has pivoted toward custom silicon while still buying Nvidia. So in one sense, Big Tech is trying to catch up. But GTC 2026 will still reveal how far behind they are: Nvidia’s data center revenue has reached tens of billions per quarter, and the company continues to set the pace for performance, software, and ecosystem. Custom chips from hyperscalers are growing, but JPMorgan has projected they will account for 45% of the AI chip market by 2028, up from 37% in 2024—meaning the majority of AI accelerator demand will still flow through merchant vendors, with Nvidia dominant. GTC is where Nvidia shows the next generation of that dominance, and where the gap between Nvidia and everyone else is put on display.
The conference is thus a referendum on dependency. If the biggest cloud and AI companies could match or replace Nvidia at scale, they would not need to show up. But they do show up, because the reality is that Nvidia remains the backbone of AI infrastructure for most of the industry. As 24/7 Wall St put it, Nvidia has cemented its role as that backbone. GTC 2026 will reveal how far behind the rest of Big Tech is: not because they are not trying, but because catching up to years of CUDA, optimized libraries, and ecosystem lock-in is a multi-year project. The keynote will showcase the next turn of the wheel. The real story is that the rest of Big Tech is still turning to get there.
Why can’t Big Tech just replace Nvidia?
Replacing Nvidia at scale is hard for several reasons. First, Nvidia’s CUDA ecosystem has millions of developers and deep integration with frameworks like PyTorch and TensorFlow; switching to custom silicon often requires rewriting or re-optimizing code and retraining engineers. Second, Nvidia’s libraries and software stack represent years of performance tuning; in-house chips must match or beat that performance while also justifying the R&D cost. Third, supply and lead times: Nvidia ships at massive scale, while hyperscalers’ custom chips are still ramping. So even as Amazon, Google, and Microsoft invest heavily in Trainium, TPU, and Maia, they still buy huge volumes of Nvidia GPUs. GTC 2026 is where Nvidia shows the next generation of that offering, and where the gap between “we are building our own” and “we have replaced Nvidia” is made visible. Big Tech is behind, and the conference is the proof.
What This Actually Means
The real story of GTC is not Nvidia’s next chip, but how dependent its biggest customers have become on a single vendor. GTC 2026 will reveal how far behind the rest of Big Tech is on AI infrastructure: they are building their own chips and reducing reliance, but the majority of AI compute still runs on Nvidia, and the gap is still wide. Until that changes, every GTC is a reminder of who sets the pace.
What is the AI infrastructure market and who dominates it?
The AI infrastructure market includes the hardware, software, and data center capacity used to train and run large AI models. It has been estimated at over $200 billion globally, with spending concentrated among a small set of players. Nvidia dominates the GPU and accelerator segment, with roughly 80% to 90% of the AI accelerator market by revenue. Microsoft, Amazon, Google, and Meta are among the largest buyers of Nvidia chips and also operate the biggest public clouds. So the market is both a chip story (Nvidia) and a cloud story (hyperscalers). GTC matters because it is where Nvidia sets the technical and product agenda for that chip layer. When Big Tech shows up, it is partly to see what they are still dependent on and how far they have to go to change that.
Sources
TechCrunch, 24/7 Wall St, Yahoo Finance, Yahoo Finance, Wedbush, The Meridiem