> ## Content Index > Fetch the complete content index at: https://www.notatechguy.com/llms.txt > Use this file to discover other available public pages before exploring further. # OpenAI claims ChatGPT speeds up CAD and PCB design - URL: https://www.notatechguy.com/openai-claims-chatgpt-speeds-up-cad-and-pcb-design/ - Published: 2026-11-04T06:18:18.000Z - Updated: 2026-11-04T06:18:18.000Z - Description: OpenAI claims ChatGPT and Codex accelerate CAD, PCB design and simulation, but offers no benchmarks or customer names to back it up. - Author: Marcello Babbili - Tags: Technology & AI, OpenAI, Google OpenAI claimed on November 3, 2026 that ChatGPT and Codex now help engineers accelerate CAD and PCB design across the physical world [S¹](https://openai.com/index/accelerating-engineering-for-the-physical-world?ref=notatechguy.com)[P³](https://openai.com/index/accelerating-engineering-for-the-physical-world/?ref=notatechguy.com). The company describes capabilities spanning mechanical and electrical engineering through to end-to-end system design, and says its own internal teams already use them to prototype concepts and cycle through design revisions more quickly [P³](https://openai.com/index/accelerating-engineering-for-the-physical-world/?ref=notatechguy.com). What the announcement does not contain: a single benchmark, a named customer, or any third-party verification. **My read:** This is a marketing announcement dressed as a capability update. OpenAI is making a broad claim, that its language models can accelerate specialised engineering workflows like PCB layout and CAD simulation, without publishing a single number to support it. I would watch for whether any hardware company or CAD vendor corroborates this with a concrete before-and-after. Until then, "helping engineers" is a claim, not a result. The gap between the claim and the evidence is the story. OpenAI says its models help engineers "build physical products" across mechanical and electrical engineering [P³](https://openai.com/index/accelerating-engineering-for-the-physical-world/?ref=notatechguy.com)[S¹](https://openai.com/index/accelerating-engineering-for-the-physical-world?ref=notatechguy.com). The language is expansive. The proof is absent — no error bars, no released benchmark code, and the acceleration figure is the vendor's own. No independent third party has verified that ChatGPT or Codex accelerate CAD and PCB design workflows. No quantitative benchmarks appear in the announcement. No customer is named. The only corroboration in the evidence pack is a Google News aggregator snippet [S²](https://news.google.com/rss/articles/CBMifkFVX3lxTE1CS3RYM1JDQWZNbmRHUXB0SEY5ZkxoWGc3ZVZ2Rmw2YzBNeTZEd3RKWEk1NzEwN0l0NG01TzV5SXBGdkQ1emJHdzRWUFdtTUpfdmFXcWFNRVJuVDlhOFhzeHd0TkQ5NXJzeVRUYndHZGtZYkMxZnp3bVdFZzlVdw?oc=5&ref=notatechguy.com), which algorithmically grouped the story alongside NVIDIA and Google headlines but contains no independent reporting. ### OpenAI's self-reported internal use The primary source confirms the announcement was published November 3, 2026 and covers advances in how OpenAI models assist with physical product design [P³](https://openai.com/index/accelerating-engineering-for-the-physical-world/?ref=notatechguy.com). OpenAI reports that its own engineers are using the models to prototype concepts and work through design iterations more quickly [P³](https://openai.com/index/accelerating-engineering-for-the-physical-world/?ref=notatechguy.com) — a self-reported claim about OpenAI's internal teams, not external customers. That distinction matters. Internal teams have direct access to OpenAI engineers, custom integrations, and models tuned for their specific tasks. A PCB designer at a hardware startup using ChatGPT through a standard web interface faces a different proposition. The distance between "our internal teams find this useful" and "engineers everywhere can accelerate CAD work" is where the claim needs evidence it does not yet have. The broader context is competitive. AI models have moved from text generation into code, and now OpenAI is pushing into the physical engineering layer. Google DeepMind has been building world models for physical simulation and shared architecture details when it announced its Genie 4 model in late October. OpenAI's announcement comes with none of that specificity. ### Testing the claim against real engineering work A mechanical engineer or PCB designer reading this should treat the claim as untested. The practical first step: take one real workflow, a PCB layout review or a CAD parameter sweep, and run it through ChatGPT or Codex with concrete inputs and constraints. Compare the output against what a human engineer produces in the same time. If the model cannot handle the file formats or the tolerance calculations that real engineering requires, the acceleration claim does not hold for that workflow. For a PCB designer at a mid-size electronics firm, that means the daily layout review and design-rule check cycle stays manual until a vendor or peer publishes a reproducible benchmark. OpenAI's open-weight models, gpt-oss-120b and gpt-oss-20b, are available under Apache 2.0 on GitHub with 20,495 stars against 2,188 forks [P⁵](https://github.com/openai/gpt-oss/?ref=notatechguy.com), and could in theory be fine-tuned for engineering-specific tasks. But that is a research project, not a product, and no evidence in the pack shows either model has been applied to CAD or PCB work. The next checkpoint is whether a hardware company or engineering tool vendor such as Autodesk, Altium or Cadence confirms an integration or publishes a benchmark. Until one does, OpenAI's claim rests on its own word. --- *Sources: [S1 — Accelerating engineering for the physical world](https://openai.com/index/accelerating-engineering-for-the-physical-world?ref=notatechguy.com) · [S2 — Accelerating engineering for the physical world - OpenAI](https://news.google.com/rss/articles/CBMifkFVX3lxTE1CS3RYM1JDQWZNbmRHUXB0SEY5ZkxoWGc3ZVZ2Rmw2YzBNeTZEd3RKWEk1NzEwN0l0NG01TzV5SXBGdkQ1emJHdzRWUFdtTUpfdmFXcWFNRVJuVDlhOFhzeHd0TkQ5NXJzeVRUYndHZGtZYkMxZnp3bVdFZzlVdw?oc=5&ref=notatechguy.com) · [P3 — Accelerating engineering for the physical world | OpenAI](https://openai.com/index/accelerating-engineering-for-the-physical-world/?ref=notatechguy.com) · [P4 — huggingface/accelerate](https://github.com/huggingface/accelerate/?ref=notatechguy.com) · [P5 — openai/gpt-oss](https://github.com/openai/gpt-oss/?ref=notatechguy.com) · [P6 — projects/chatgpt.yaml](https://github.com/opening-up-chatgpt/opening-up-chatgpt.github.io/blob/main/projects/chatgpt.yaml?ref=notatechguy.com)* --- *Written from 6 sourced items, 5 of them primary.* ## More from Not A Tech Guy - [FedReady AI agents find all federated learning sites at 65% precision](https://www.notatechguy.com/fedready-ai-agents-find-all-federated-learning-sites-at-65-precision/) - [Prompt injection defense cuts agent attack rate from 37.8% to 4.4%](https://www.notatechguy.com/prompt-injection-defense-cuts-agent-attack-rate-from-37-8-to-4-4/) - [PR-Agent trends on GitHub with 122 new stars in a day](https://www.notatechguy.com/pr-agent-trends-on-github-with-122-new-stars-in-a-day/)