--- title: Extropic tapes out Z1 thermodynamic chip, claims 10,000x... description: Extropic has taped out its Z1 chip, a thermodynamic computing processor running 269,000 probabilistic bits on under 1W of power. The company claims 10,000x... canonical: https://daily.dev/posts/extropic-tapes-out-z1-thermodynamic-chip-claims-10-000x-better-energy-efficiency-than-gpus-kynmpb80v twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Extropic tapes out Z1 thermodynamic chip, claims 10,000x better energy efficiency than GPUs | daily.dev og:description: Extropic has taped out its Z1 chip, a thermodynamic computing processor running 269,000 probabilistic bits on under 1W of power. The company claims 10,000x... og:url: https://daily.dev/posts/extropic-tapes-out-z1-thermodynamic-chip-claims-10-000x-better-energy-efficiency-than-gpus-kynmpb80v og:image: https://api.daily.dev/og/posts/KynMPb80v.png og:image:alt: Extropic tapes out Z1 thermodynamic chip, claims 10,000x better energy efficiency than GPUs og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Extropic tapes out Z1 thermodynamic chip, claims 10,000x better energy efficiency than GPUs **[Collections](https://daily.dev/sources/collections)** · 1 min read · 0 upvotes · 0 comments ## Summary Extropic has taped out its Z1 chip, a thermodynamic computing processor running 269,000 probabilistic bits on under 1W of power. The company claims 10,000x better energy efficiency than GPUs for probabilistic workloads, with the chip now in fabrication. Alongside the hardware, Extropic released a PyTorch-equivalent library to ease developer adoption. Thermodynamic computing has historically faced skepticism — using thermal noise as a computational resource rather than suppressing it has rarely succeeded — but reaching fabrication moves the technology from theory to silicon. ## Content Extropic has taped out its Z1 chip, a thermodynamic computing processor that runs 269,000 probabilistic bits on less than 1W of power. The company claims it's 10,000x more energy efficient than GPUs for probabilistic workloads, and the chip is now in fabrication. Alongside the hardware, Extropic released a PyTorch-equivalent library for thermodynamic hardware, which should lower the barrier for developers to actually use it. Thermodynamic computing has a long history of skepticism. The core idea - using thermal noise as a computational resource rather than fighting it - has been attempted before without much success, and plenty of experts have dismissed it as a dead end. Extropic has been drawing that criticism since day one. But getting a chip into fabrication is a real milestone. Whether the efficiency claims hold up at scale is another question, but Z1 at least moves the conversation from theory to silicon. If it works anywhere close to spec, it'd be one of the more surprising hardware stories in recent memory. ## Community take How the wider developer community reacted, aggregated from 3 discussions and 23 comments across x (as of 2026-08-05). **TL;DR:** The community is cautiously intrigued but skeptical — the 10,000x efficiency claim draws both SpaceX-style optimism and outright dismissal as grift, with the central question being whether the advantage survives real-world system overhead and whether today's transformer workloads can even map onto probabilistic hardware. **Sentiment:** 35% positive · 40% mixed · 25% skeptical **The case for** - Reaching tape-out moves the technology from theory to actual silicon, lending it more credibility than prior thermodynamic computing efforts. - The PyTorch-compatible library is seen as a critical unlock, since developers won't build on hardware they can't program. - Some draw parallels to early LLM skepticism, arguing experts have been wrong before about paradigm-shifting compute approaches. **The pushback** - The 10,000x efficiency claim is unverified and needs to survive full-system overhead including noise handling, interfacing, and software maturity. - It's unclear whether today's transformer-based workloads can map cleanly onto probabilistic bits, limiting near-term applicability. - At least some commenters flatly dismiss the company as delusional grifters with unproven technology. - Transitioning away from deterministic digital logic requires rewriting entire ML algorithm and software stacks, a massive undertaking. **By community** - x (mixed): Replies split between SpaceX-style optimism and outright skepticism, with the most substantive technical concern being whether efficiency gains survive real-world system overhead and whether current transformer workloads can exploit probabilistic hardware. **Hottest debate:** Whether the 10,000x efficiency advantage is a genuine breakthrough or an unsubstantiated claim that will collapse under full-system overhead and software immaturity. **Open questions** - Does the efficiency advantage hold once noise handling, interfacing, and software stack overhead are factored in? - Can today's transformer-based models map cleanly onto probabilistic bits, or does the chip only pay off once models are redesigned around noise? - What is the path to general programmability beyond probabilistic sampling? - How well does the chip handle hardware noise variability in practice? **Highlights** > @rohanpaul_ai Probabilistic silicon finally matches how models actually run: sampling, not determinism. The real question: does it help inference on today's transformer stack, or only pay off once we build models around noise instead of fighting it? > — [vikasmalpani on x · 3 points](https://x.com/vikasmalpani/status/2084902655445680423) > @rohanpaul_ai Ambitious efficiency claims at tape-out. The real test is whether the advantage survives full-system overhead (noise handling, interfacing, software maturity) and whether today’s transformer workloads actually map cleanly onto probabilistic bits. > — [FerryLee\_AIPOCH on x](https://x.com/FerryLee_AIPOCH/status/2084909819287495028) > @rohanpaul_ai the pytorch frontend is the actual unlock here, 10k x efficiency claims need silicon proof but nobody builds on hardware they can't program > — [ShinkaIoT on x](https://x.com/ShinkaIoT/status/2084890083724054913) > @Hesamation They are delusional grifters bro > — [PriestruYuuru on x · 2 points](https://x.com/PriestruYuuru/status/2084994099099427230) > @Hesamation Transitioning away from deterministic digital logic requires rewriting machine learning algorithms and software stacks to natively process uncertainty; and is the most ambitious project I have seen.. I would love to see what algorithms that they have used in house.. I have > — [volatilemarkts on x](https://x.com/volatilemarkts/status/2085052851663917462) **Source threads** - [x](https://x.com/rohanpaul_ai/status/2085111401169993815) · 0 points · 0 comments - [x](https://x.com/Hesamation/status/2084887216728191212) · 1 points · 17 comments - [x](https://x.com/rohanpaul_ai/status/2084887123480793530) · 0 points · 6 comments ## Similar posts on daily.dev - [Extropic](https://daily.dev/posts/extropic-6efhaf2wc) · Hacker News · 0 upvotes · 0 comments - [Researchers Propose Thermodynamic Computing Architecture That Could Dramatically Reduce AI Energy Use](https://daily.dev/posts/researchers-propose-thermodynamic-computing-architecture-that-could-dramatically-reduce-ai-energy-us-yrq4oxapf) · The Quantum Insider · 1 upvotes · 0 comments --- Tags: [#hardware](https://daily.dev/tags/hardware), [#pytorch](https://daily.dev/tags/pytorch) [View this post on daily.dev](https://daily.dev/posts/extropic-tapes-out-z1-thermodynamic-chip-claims-10-000x-better-energy-efficiency-than-gpus-kynmpb80v) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Extropic tapes out Z1 thermodynamic chip, claims 10,000x better energy efficiency than GPUs","url":"https://daily.dev/posts/extropic-tapes-out-z1-thermodynamic-chip-claims-10-000x-better-energy-efficiency-than-gpus-kynmpb80v","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/extropic-tapes-out-z1-thermodynamic-chip-claims-10-000x-better-energy-efficiency-than-gpus-kynmpb80v"},"datePublished":"2026-08-05T21:11:28.481Z","dateModified":"2026-08-05T21:12:14.112Z","description":"Extropic has taped out its Z1 chip, a thermodynamic computing processor running 269,000 probabilistic bits on under 1W of power. The company claims 10,000x...","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/extropic-tapes-out-z1-thermodynamic-chip-claims-10-000x-better-energy-efficiency-than-gpus-kynmpb80v","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"hardware,pytorch","timeRequired":"PT1M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Extropic tapes out Z1 thermodynamic chip, claims 10,000x better energy efficiency than GPUs"}]} ```