Silicon tales

The human story behind the silicon. We make hardware read like drama, not a datasheet. GPUs, new architectures, and retro machines.

[Silicon tales @talessilicon]

Nebius grew revenue 684% in a single year, signed $27 billion with Meta, and is now priced as though the trillion-dollar outcome is a matter of time. The filings tell a more interesting story.

https://t.co/stqgIV5nSY

[Silicon tales @talessilicon]

Kimi-K2 just been released!

[Silicon tales @talessilicon]

The AI factory blueprint: match the silicon to the strategy...
Anyone can read a spec sheet. The harder question is which NVIDIA system your strategy actually needs, and at which numeric precision. Read the article:
https://t.co/Gyg8vkeSBx

[Silicon tales @talessilicon]

GPT-5.6 sol launches tomorrow. Stay tuned.

[Silicon tales @talessilicon]

Running an LLM locally: how it works, how much room it needs, and which machine it fits.

https://t.co/Yzuir7U1Wx

[Silicon tales @talessilicon]

Stopping the bad guys with Cloudflare: 219 malicious requests blocked or challenged in the last month #cloudflare

[Silicon tales @talessilicon]

OpenAI just announced their flagship GPT model.

[OpenAI @OpenAI]

Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model, as well as GPT-5.6 Terra, a balanced model for efficient, everyday work, and GPT-5.6 Luna, a fast and affordable model for high-volume work.

https://t.co/OoM83SyISN

[Silicon tales @talessilicon]

How the Commodore 64 Conquered the Bedroom, and Why Its Maker Did Not Survive It
The best-selling computer ever made was a beige box with a red light. This is the story of the survivor who built it as a weapon, the engineers who taught it to sing.

https://t.co/Fq0dpapQLm

[Silicon tales @talessilicon]

Just announced: Sakana Fugu is a full multi agent orchestration system through a single API, dynamically coordinating frontier models for complex tasks with frontier level performance and no single vendor lock in.
https://t.co/6Qc4MXaY4K

[Silicon tales @talessilicon]

Two Turkish Engineers Built the First GPU. History Gave the Credit to Someone Else.

This is the story of 3Dlabs, and of the two men almost nobody remembers.

https://t.co/OUpPeeKh5v

[Silicon tales @talessilicon]

Microsoft just launched Copilot Cowork: an AI coworker across Outlook, Teams, Word, Excel, and more. But it’s not just a subscription—companies also pay for usage. We went from buying software once, to SaaS, to pay-per-AI.

[Silicon tales @talessilicon]

Unreal Engine 5.8 is now live!
What’s new in Unreal Engine 5.8 👇
https://t.co/9OnkXys2Pp

[Silicon tales @talessilicon]

Good old days.
When a 3dfx Voodoo upgrade didn’t just give you more FPS, it gave you a whole new personality. Quake, Quake 2, Unreal, Sin, Turok, Hexen II… peak PC gaming era.

[Silicon tales @talessilicon]

Apple’s upcoming camera-equipped AirPods.

[Silicon tales @talessilicon]

See thread for details on how GLM-5.2 (Max) performs across 5 different signals.

[ZCode @zcode_ai]

GLM-5.2 is here: stronger reasoning, advanced coding, and solid 1M-token context for long-horizon engineering tasks.
Paired with ZCode, it delivers a smoother, more seamless agentic coding experience across debugging, refactoring, project understanding, and multi-step execution.
New users can start with the ZCode Start Plan: 5 days of GLM flagship model usage, up to 5M tokens per day.
Our Linux version is also now available for testing in the Discord community.
Join our Discord community: https://t.co/oK1pV02Prf

[Silicon tales @talessilicon]

GLM-5.2 from @Zai_org is now live on Novita.

[Silicon tales @talessilicon]

GLM-5.2 is now 1st on Design Arena.
@Zai_org

[Silicon tales @talessilicon]

SpaceX is acquiring Cursor in an all-stock deal valuing the company at $60B, with the merger expected to finalize in Q3 2026.

[Silicon tales @talessilicon]

Fable 5 access might be restored within the next 72 hours.