As AI is increasingly used to automate tasks and answer queries, Nvidia's graphics processors face growing competition from central processors and custom chips that are better suited to handle that process — known as inference.
The shift has prompted major technology companies to invest in their own silicon.
Meta plans to begin manufacturing its in-house "Iris" AI chip in September, Reuters exclusively reported, part of a four-generation custom silicon project designed to lower computing costs.
Alphabet has ordered over 3 million chips from Intel for 2028, the Information reported in June, adding that Nvidia is also evaluating the American chipmaker's manufacturing technology to make a processor that combines four GPUs into a single unit.
Rivals Intel and AMD are also targeting the inference market and several Chinese firms including Baidu already produce their own chips for such tasks.
In March, Nvidia unveiled a new central processor and AI system based on technology licensed from inference-focused startup Groq in a $17 billion deal, pairing Groq's chips with its upcoming Vera Rubin platform.
The company's dominance as the AI market leader was reinforced earlier this month when SpaceX CEO Elon Musk said the rocket firm would exclusively use Nvidia's hardware.
Nvidia has said the revenue opportunity for its AI chips could exceed $1 trillion through 2027, doubling from the $500 billion opportunity through 2026 that Nvidia cited for its Blackwell and Rubin chips.
Nvidia's second-quarter revenue more than doubled to $96.22 billion, beating estimates of $92.17 billion. Adjusted profit was $2.22 per share for the three months ended July 26, compared with estimates of $2.10.
Data center revenue also more than doubled to $89 billion, exceeding estimates of $85.08 billion.
The $5.16 trillion company has surged past Wall Street's sales expectations over the past 15 quarters, but the magnitude of the beats has narrowed.
NVDA•
