AIToday
Large Language ModelsSemafor TechPublished: Aug 27, 2026, 04:01 JST2 min read

OpenAI chip rivals Nvidia in inference speed

OpenAI chip rivals Nvidia in inference speed

Key takeaway

  • OpenAI's new inference chip, Jalapeño, matches Nvidia's best performance.

  • OpenAI used its own AI to design it quickly.

  • This could accelerate AI development and empower smaller firms.

3 Key Points

  1. What happened

    OpenAI unveiled benchmarks for its Jalapeño inference chip on Tuesday, showing it competes with Nvidia’s best chips. The chip was developed in record time using OpenAI’s own AI models to accelerate design and verification.

  2. Why it matters

    This marks a shift where AI companies can use their own models to develop chips, which then run more powerful models to develop the next chip—a self-reinforcing cycle that could speed up capability growth. It also signals that smaller companies may become more ambitious and challenge larger competitors, spreading beyond software into physical industries.

  3. What to watch

    OpenAI still relies on Broadcom for building and fabs like TSMC for manufacturing, but the ability to design rapidly could reshape how chips are created. The flywheel effect may lead to even faster acceleration, though bottlenecks in memory, power, and limited fab capacity remain.

Ask the AI about this article →

Context & Analysis

OpenAI's announcement signals that AI firms are no longer just users of chips but potential designers. By leveraging its own models to accelerate the design and verification process, OpenAI has shown that AI can be a tool for creating the very hardware it runs on. This could shorten development cycles drastically, enabling faster iterations of both models and chips.

The ability to move from concept to competitive chip quickly may also lower barriers for smaller companies. They can adopt similar AI-driven design approaches to challenge incumbents in various industries. While the immediate impact is in software and chip design, the potential for disruption in physical products is noted in the article.

However, significant bottlenecks remain—memory, power, and limited fab capacity. The article suggests the market will resolve these over time, but it may temper the speed of acceleration. The reliance on Broadcom and TSMC also underscores that even with faster design, manufacturing and integration still require established partners.

FAQ

How fast is OpenAI's Jalapeño chip compared to Nvidia's?
The benchmarks show it is so fast that it competes with Nvidia's best chips, according to OpenAI's Tuesday announcement.
Who helped OpenAI build the chip?
OpenAI worked with Broadcom to build it and will rely on fabs like TSMC for manufacturing.
What makes Jalapeño different from other chips?
It was developed in record time by using OpenAI's own AI models to speed up the design and verification phase, a process that could create a flywheel for faster chip development.

Get the latest Large Language Models news every morning

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · takes 30 seconds · unsubscribe anytime

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleNew benchmark shows AI learning speed doubles every 3 months