Nvidia's record earnings ignite AI chip boom, boosting rivals Broadcom, AMD and storage players
Nvidia posted a record $96.2 billion of revenue in its fiscal second quarter ended July 26, 2026, up 106% from a year earlier and ahead of the $92.2 billion consensus. The print has pulled attention onto the wider AI hardware complex, where Western Digital, SanDisk and Super Micro all reported AI-driven revenue growth. Two figures circulating alongside the story need care: Broadcom's $63.89 billion is its already-reported fiscal 2025 full-year total, not a fresh quarter, and OpenAI's Jalapeno accelerator claims up to 1.9x better efficiency, with the larger 3.6x figure referring to latency reduction.
NVDA reported a record $96.2 billion of revenue for its fiscal second quarter ended July 26, 2026, up 106% from a year earlier and roughly 18% sequentially, beating the $92.2 billion consensus. Shares have not simply run on the print: NVDA last traded at $217.55, down 4.6% on the session, and SentiSense's own read has the stock at Strong Bullish but cooling, with the 30 day average score at 42.3 against a reading of 23.2 today.
Two numbers moving with this story need to be labelled carefully. The $63.89 billion figure attached to AVGO, up 24%, is Broadcom's full fiscal 2025 revenue, disclosed in December 2025. It is not a fresh quarterly result reported alongside Nvidia. Broadcom's next report is a separate event: fiscal third quarter results are company-confirmed for September 2, 2026 after the close, with consensus around $29.4 billion of revenue. Reading the annual total as this week's quarter would badly overstate the run rate.
The read-through into storage and servers is the more durable part of the story. Western Digital grew revenue 44% year over year, SanDisk 372%, and Super Micro Computer 93%, all on AI-driven demand. One caveat the bullish framing tends to skip: Super Micro's shares are down about 15% over the past year despite that revenue growth, so the case there is forward-looking rather than a description of what the tape has already done.
OpenAI's in-house Jalapeno accelerator is the other item worth restating precisely. The published comparisons against Nvidia's GB200 and GB300 parts show up to 1.9x better efficiency and up to 3.6x lower latency . The 3.6x number is a latency figure, not an efficiency figure, and the two are frequently merged. Similarly, the price competition coming out of China is happening in AI model and API pricing rather than in chip hardware, so it may compress the economics of inference services before it touches semiconductor margins. Watch Broadcom's September 2 guidance and Nvidia's next data center commentary for whether the demand picture holds.
Related Stocks
Powered by SentiSense - Intelligent Market Analysis