Google Gemini 3 launch leaked

PLUS: OpenAI bets big on custom chips to scale ChatGPT and Sora

Author

Barsee
14 Oct

Together with

Howdy, it’s Barsee.

Happy Tuesday, AI family, and welcome to another AI Valley edition. This issue takes 6 minutes to read.

Today’s climb through the Valley reveals:

  • Google Gemini 3 launch leaked

  • OpenAI bets big on custom chips to scale ChatGPT and Sora

  • Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

WISPR FLOW

Image Credit: Flow

AI has changed how we think, but not how we communicate. Wispr Flow bridges that gap, letting you speak naturally while it formats and edits in real time. It understands context, tone, and corrections, so "actually, make that 6 p.m." updates itself on the spot.

It's not just for writing faster, it's for freeing up the mental space that typing steals. Once you start using Flow, the keyboard starts to feel... optional.

*This is sponsored

THROUGH THE VALLEY

Leaked company material

A leaked internal memo suggests Google will launch Gemini 3.0 on October 22, aligning with a “marketing milestone” noted in company materials. The update reportedly brings big jumps in multimodal reasoning, coding, latency, and even music generation (see example), along with companion releases like Veo 3.1 and a Nano Banana variant of Gemini 3 Pro.

Meanwhile, some developers appear to have early access through an “ecpt checkpoint” labeled Gemini 3.0 Pro in AI Studio A/B tests. One red-team tester known as Chetaslua claimed the model built a fully functional, macOS-style web operating system, complete with a text editor, paint app, video editor, and working browser, in a single HTML file. The post quickly went viral, with the author clarifying: “not cherry-picked, Google didn’t pay me.”

Credit: Chetaslua

Why does it matter?

If authentic, Gemini 3 may be one of Google’s most significant AI progress since Gemini 1’s debut, merging reasoning, coding, multimedia generation, and autonomous task handling into one architecture. Industry watchers note that early demos already blur the line between standalone applications and AI-native operating environments, raising questions about how “general-purpose AI” will be defined going forward. With OpenAI’s rumored “Q*” model and Anthropic’s Claude 4.5 in the same competitive window, Gemini 3’s release could ignite a new phase in AI platform wars.

OpenAI has entered a multi-year partnership with Broadcom to co-design and deliver 10 gigawatts of custom AI accelerators and networking systems between 2026 and 2029, forming the digital backbone for OpenAI’s growing set of models and global data centers. This capacity equals the electricity usage of over eight million U.S. homes and represents one of the largest single investments in chip infrastructure ever attempted.

The companies have been quietly collaborating for 18 months, designing not just silicon but full rack-level systems optimized for OpenAI’s workloads. OpenAI also leveraged its own AI models to accelerate the chip design process, cutting months off the typical engineering cycle.

CEO Sam Altman called the partnership “a critical step to unlock AI’s potential,” making it OpenAI’s third megadeal in weeks, alongside recent agreements with Nvidia and AMD as part of Stargate.

Why does it matter?

Owning custom silicon and the full hardware-software stack is meant to align everything (from chips to developer tools) under OpenAI control, mirroring how Apple transformed consumer electronics and how Microsoft built the PC ecosystem. The new chips and systems will power not just AI models but also consumer hardware and developer platforms in OpenAI’s bid for vertical integration of the entire AI experience.

TRENDING TOOLS

  • Wand > Draw anything. Render anything. Your imagination, visualized instantly

  • Nano Banana > Now built into Google Search and NotebookLM and coming soon to Photos

  • Microsoft MAI-Image-1 > Microsoft’s first in-house text-to-image generator, capable of producing stunning, photorealistic visuals

  • Google AI Studio > The latest update lets you “vibe code” using your voice

  • StreamingVLM > Real-Time understanding for infinite video streams

THINK PIECES / BRAIN BOOST

THE VALLEY GEMS

What’s trending on social today:

THAT’S ALL FOR TODAY

Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee

👍️ New reader? Subscribe here

Thanks for being here.

REACH 100K+ READERS

Acquire new customers and drive revenue by partnering with us

Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.

If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.