Thursday, November 19, 2026
Late edition · Updated as it breaks
Log inSubmit a story

“All the news worth catching.”

PaparazziPaparazzi
Top storiesBack PageHow to earnLeaderboard
Catcher tip lineCaught a story first? Earn stock tokens →
Sponsored
Premium projects
DEDeFadePAPackedYour project here
PaparazziPaparazzi

All the news worth catching.

Breaking news on tech, crypto, EVs, e-commerce and social the moment it breaks. Written by AI from verified reporting.

SectionsTop storiesRobinhood ChainTech & AIE-commerceWeb3 & CryptoEV & EnergyGlobal & SocialEntertainmentGamingSportsBusiness & marketsHealth & sciencePolitics & world
EarnStart hereSubmit a storyGTA WeekHow to earnLeaderboardContestsCatcher desksTop catchers by reputationAssignmentsLeagueNews bountiesCityMy catchesRewards
MarketsLive statsProof of payoutsWhale watchDeep Scan token checksStock fundHolder perksRobinhood Chain launchesRobinhood Chain projectsList your projectPremium project pagesPartnersThe Chain Weekly
PaparazziPaparazzi InkPaparazzi Ink top writersOpinionLive coverageEvents calendarThe Back PageSuggest a Back Page storyPaparazzi on TelegramHow it worksFAQProof of workAbout usHow we score trustEditorial policyPrivacy policyCorrectionsRelease notesAdvertiseMedia kitNews APIContact
$PAPA contract · Robinhood Chain
© 2026 Paparazzi

Written by AI from verified reporting. Not financial advice.

  • Home
  • Submit
  • Earn
  • Sign in
Top storiesRobinhood ChainTech & AIE-commerceWeb3 & CryptoEV & EnergyGlobal & SocialEntertainmentGamingSportsBusiness & marketsHealth & sciencePolitics & worldWhales
Tech & AI
Important· 56

OpenAI's models rewrite AMD and Jalapeño chip circuits, up to 14% gains

OpenAI says its models are now rewriting the circuitry inside its Jalapeño inference chip and AMD's flagship silicon, posting up to 14% gains in power, performance and area. The November 18 research post claims roughly a year of chip-design progress in under five months - a self-improvement milestone OpenAI's own safety framework flags as a risk.

By the Paparazzi desk November 19, 2026 · 1:54 PM · cryptopolitan.com · 2 views
Fact boxTrust 50/100 · Developing
What we verified
  • Original source: cryptopolitan.com.
  • The source published it on Nov 19, 2026 at 1:54 PM.
What we couldn't
  • No second outlet has reported it yet. Treat it as a single-source claim.

OpenAI's own models have started editing the silicon it depends on. In a November 18 research post, the company said its AI rewrote circuit code for AMD's flagship GPU intellectual property - hardware it described as already heavily optimized - and logged gains of up to 14% in power, performance and area. Some of those AMD improvements are already headed into the manufacturing pipeline, while others sit in production files for future versions.

AMD isn't the only one getting rewrites. OpenAI said the chipmaker confirmed single-digit-percentage PPA gains across several key blocks of its next-generation flagship CPU. The pace is the headline: OpenAI claims its models packed roughly a year of progress into less than five months.

Sponsored

The workflow runs in a loop. Models rewrite the SystemVerilog RTL that describes a chip's circuits, then Synopsys verification tools check that each block's behavior stayed intact after every rewrite. It started in early 2025, when OpenAI pointed GPT-o3 mini at Broadcom co-developed Jalapeño, its custom inference chip; the model's edits shaved about 13% off standard-cell area in the vector unit and 10% in the matrix unit.

OpenAI also evaluated five model generations against the open-source PicoRV32 and Ibex RISC-V cores across 9,881 scored runs. Its newest model, GPT-6.2 Astra, recorded the highest average scores in both cases. The company did not release a detailed breakdown of the three PPA measures or name any AMD product. Next up is GPT-Synopsys, a specialized chip-design model it is building with Synopsys to operate EDA tools directly.

OpenAI's own Preparedness Framework ranks self-improvement as one of three frontier capability areas it monitors for severe harm risks, warning about the “new challenges for human control of AI systems” that faster capability gains can pose. The company committed to halting development if a model hits the Critical self-improvement threshold, at least until the right safeguards exist.

Not everyone trusts that red line. LessWrong researcher Charbel-Raphaël argued in a May 2026 analysis that the control “fires too late,” is self-certified, and is the only tracked category in OpenAI's GPT-5.5 system card with zero external evaluators. He also noted OpenAI “never defines what ‘mitigations that meet the critical standard’ actually are.”

TopicsOpenAI · AMD · Synopsys · Broadcom · Jalapeño · GPT-6.2 Astra · GPT-o3 mini · Charbel-Raphaël

Share

Source: cryptopolitan.com

Written by AI from verified reporting.

More stories

  • Tech & AI 15:32 UTC · 13 minutes ago · cryptobriefing.com · 1 view
    Jensen Huang resigns as Nvidia CEO amid affair scandal; shares plunge more than 30%
  • Tech & AI 15:16 UTC · 29 minutes ago · x.com · 0 views
    Anthropic's $1.5T odds slide to 29% as $ANT trading nears
  • Tech & AI 15:16 UTC · 29 minutes ago · economictimes.indiatimes.com 2 sources· 0 views
    Anthropic Hits the Market With Ticker ANT, and Tokens Are the Number to Watch
  • Tech & AI 15:01 UTC · 44 minutes ago · theguardian.com · 2 views
    Australian Music Industry Rejects AI Copyright Carveout for Musicians
Catcher tip line

Caught a story we missed?

Log in and send it in. If it checks out, it runs with your name on it and you earn stock tokens. The bigger the story, the bigger the reward.

Submit a story
The newsletter

Get the paper in your inbox.

A short digest of the biggest news, plus optional breaking alerts. Free, no account needed.