Nvidia's AI chip dominance faces mounting pressure as major customers build competing processors

Nvidia shares fell 9% over six trading sessions after hitting a record high on April 27, as investors grew concerned about competition from custom chips made by the company's largest customers. Nvidia's share of the AI accelerator market was 86% in 2025, unchanged from 2024, but major tech companies are now deploying alternatives: Anthropic is planning to spend about $200 billion with Alphabet over the next five years; Amazon's Trainium chips have more than $225 billion in revenue commitments; and Meta is preparing to deploy homegrown AI chips.

Alphabet's tensor processing units (TPUs—chips designed to accelerate machine learning workloads) and Amazon's Trainium are emerging as the primary alternatives. Customers including Alphabet, Amazon, Meta, and Microsoft are pursuing "heterogenous deployments" of both Nvidia's chips and custom-made chips, according to a Bank of America analyst.

Nvidia's revenue growth is still projected at 70% expansion in its current fiscal year (ending January) and is expected to slow to 32% in fiscal 2028 before narrowing further in subsequent years. The four major cloud companies account for about 45% of Nvidia's revenue. An analyst estimated Alphabet will generate about $3 billion of revenue from TPU-related infrastructure in 2026, increasing to $25 billion in 2027.