SemiconductorsLOT 261119 · 3 min read
OpenAI Claims Up to 14% PPA Gain on AMD's Leading GPU IP
OpenAI claims up to 14% PPA gains on AMD GPU IP and AMD-verified CPU block results, but vested share incentives and a Synopsys product roadmap complicate the picture.
Key takeaways
- OpenAI claims up to 14% PPA improvement on specific blocks of AMD's leading GPU IP.
- AMD verified single-digit PPA gains on key blocks of its upcoming CPU; Synopsys tools confirmed unchanged behavior.
- October 2025 deal gives OpenAI rights to buy up to 160 million AMD shares and use six gigawatts of AMD GPUs.
- AMD Data Center revenue rose 97% to $8.546 billion last quarter; Q4 guidance near $17.0 billion.
- AMD presents at UBS and Barclays conferences on December 1 and December 9.

OpenAI says its models improved circuits in Advanced Micro Devices' leading GPU IP by up to 14% against heavily optimized baselines, according to a research post the company published this week. The gain came from rewriting the SystemVerilog code that describes the chip's circuits. AMD shares closed at $530.14, up 5.31%, for a market value of $877.42 billion, and traded at $536.99 after hours. No source ties the move to the post, and AMD issued no formal statement.
The stock is up about 148% in 2026 yet still sits roughly 18% below its $649.42 record close. Every figure in the OpenAI post comes from a partner that profits when AMD looks strong.
What does the 14% actually measure?
PPA stands for power, performance and area: how much electricity a circuit uses, how fast it runs, and how much silicon it occupies. RTL is the code engineers write to describe a circuit before it becomes a physical layout.
The up-to-14% figure is a ceiling measured on specific GPU blocks. OpenAI did not break down how much came from power, performance or area. It says nothing about how fast a finished AMD GPU will run.
OpenAI frames the work as roughly a year of progress delivered in under five months. AMD added some changes to a design it recently sent for manufacturing, a step called tape-out, and others to production files for future versions. Taped out is still far from shipping. Nobody has reported measured gains on AMD silicon.
Why the CPU result carries more weight
The strongest evidence sits elsewhere. AMD itself verified single-digit percentage PPA gains across key blocks of its upcoming leading CPU, using designs written by OpenAI's model. Synopsys tools confirmed that each rewrite left the block's behavior unchanged.
The economics are simple. A smaller block means more chips per wafer and a lower cost per chip. Lower power improves performance per watt in data centers constrained by electricity supply, which is why grid and cooling suppliers keep appearing alongside the chipmakers. Against fourth-quarter revenue guidance of about $17.0 billion, block-level gains are small. Their value lies in helping AMD ship a new product generation every year.
Who keeps the edge?
OpenAI's incentives deserve a look first. Under the October 2025 deal, OpenAI agreed to use six gigawatts of AMD GPUs. It also received the right to buy up to 160 million AMD shares at a low fixed price, with vesting partly tied to AMD price targets.
The only public AMD reaction came from CEO Lisa Su, who quoted co-founder Greg Brockman's note that some results sit in a recently taped-out design and called them fantastic work by engineers at both companies.
OpenAI is also building its own chips. It estimates its model cut standard-cell area by about 13% in the vector unit and 10% in the matrix unit of its Jalapeño inference chip.
The deeper competitive question involves Synopsys. OpenAI is training GPT-Synopsys to use Synopsys's electronic design automation software, the tools chipmakers use to design and test circuits. Once that becomes a product, any chip designer can buy it, NVIDIA included.
That would make AMD's lead a matter of months. Synopsys gets paid by every customer using the tool, depends less on any single company's product cycle, and offers broader exposure to AI-assisted chip design with a stronger claim to a lasting edge.
What the valuation says
The stronger case for AMD rests on its core business. Data Center revenue rose 97% to $8.546 billion last quarter, and a forward P/E near 30 already prices in much of that growth.
AMD presents at the UBS and Barclays conferences on December 1 and December 9. If management ties OpenAI-designed changes to a named product with a ship date, the claim gains weight. If neither presentation mentions it, the 14% figure belongs in the marketing file.
247wallst.com (Source)
More from Nathan Brooks
Show full bio
Senior reporter covering people, launches and day-to-day practice in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,111 articles