Tech
OpenAI unveils first custom AI inference chip, Jalapeño, with Broadcom - and its development was sped
OpenAI and Broadcom have unveiled "Jalapeño," their first custom AI accelerator chip designed specifically for large language model (LLM) inference.
Key takeaways
- Jalapeño cuts AI inference costs by about 50%
- OpenAI's total operational expenses ballooned to $34 billion in 2025, resulting in an operating loss of nearly $20.92 billion
- The development process relied on prior generation OpenAI models to accelerate parts of the chip design
- The Jalapeño chip may offer reassurance to private investors and public markets ahead of a heavily anticipated public offering in 2026
- The ultimate goal of the OpenAI and Broadcom partnership involves deploying gigawatt-scale data centers with Microsoft and other partners beginning in 2026
OpenAI and Broadcom have unveiled "Jalapeño," their first custom AI accelerator chip designed specifically for large language model (LLM) inference. Developed in a rapid nine-month window using OpenAI's own models to accelerate design, the ASIC reportedly cuts inference costs by approximately 50%. OpenAI has already begun testing its GPT-5.3-Codex-Spark model on the chips in a test environment. The move represents a strategic effort by OpenAI to improve its unit economics, following audited 2025 financial documents showing a $20.92 billion operating loss on $34 billion in expenses.
In their words
“this is a real performance improvement...on performance per watt and performance per dollar.”
By the numbers
- 50%
- Estimated reduction in inference costs using Jalapeño
- 9 months
- Time from early schematics to fabrication readiness
- $13.07B
- OpenAI's total revenue generated throughout 2025
- $34B
- OpenAI's total operational expenses in 2025
- $30B
- Nvidia's direct investment into OpenAI in February 2026
How it unfolded
- OpenAI and Broadcom partnership publicly announced
- Nvidia finalizes $30B investment; AWS invests $50B
- Cerebras executes its initial public offering
- OpenAI and Broadcom unveil Jalapeño custom chip
Turn stories like this into views
Ravenclip finds the Tech news, makes the video, and posts it before attention moves on.
Common questions
- What happened with Jalapeño?
- Jalapeño cuts AI inference costs by about 50%
- Where can I read the original report?
- Read the full report at venturebeat.