Sponsored by
Welcome back to the Neural Net! It's Tuesday, but instead of tacos we’re serving up a full plate of fresh AI scoops.
In today’s edition: China’s AI rise was years in the making, how Gumloop helps you automate the boring stuff, Samsung and Tesla team up, Claude agrees to EU rules, and more
▼
The Street
note: stock data as of last market close
▼
🚀 The Least Surprising Story in Tech: China’s AI Rise
TL;DR
Last week we unpacked Trump’s AI push. But here’s what no one’s saying: America’s “Big AI Moment” already happened in China, in 2017.
China’s culture and ecosystem power fast, focused innovation.
AI dominance isn’t necessarily about who has the most advanced models, it’s about who’s the most embedded.
Eight Years Later: China’s AI Plan Is Paying Off
China dropped its New Generation AI Development Plan eight years ago. And it wasn’t just a vision doc, it was a detailed, technical roadmap with three phases:
By 2020: Catch up to global leaders
By 2025: Hit world-class level in core AI tech
By 2030: Become the global AI innovation center
Fast forward to now: they’ve checked off the 2025 box. Chinese DeepSeek shook up the market earlier this year by undercutting leading US models on training and operating costs for the same performance.
Now Z.ai (formerly Zhipu) has dropped GLM-4.5—an open-source, agentic model that undercuts even DeepSeek on price. For context: DeepSeek offers similar quality as U.S. rivals while being 17x cheaper.
How Did China Catch Up So Fast?
The usual narrative? They hoard chips, copy Western models, and let government money do the heavy lifting, but that’s a surface read. The real edge is structural and cultural.
Optimism vs. anxiety: It’s a “Star Trek vs. Black Mirror” affect. In China, AI is framed as progress. In the U.S., it’s viewed by many as an existential threat. The question in China is “how fast can we deploy this?”
Coordination vs. chaos: In China, “state, market, and academia row in the same direction.” Compare that to the U.S., where everybody is busy fighting with each other.
Diffusion > end state: U.S. firms obsess over the final frontier—AGI. China bets on diffusion, weaving AI into everyday life. It’s cheaper, faster, and locks in dependence on their tech.
And here’s the kicker: the U.S. is no longer widening its lead in R&D. According to AAAS, America is still the biggest spender—but barely, and the margin is the narrowest since the ’90s. Meanwhile, China is pumping out more papers, patents, and PhDs than anyone.
What AI Dominance Really Means
ChatGPT still dominates the consumer chatbot market: 910M downloads vs. DeepSeek’s 125M. But the real AI race isn’t about who hits AGI first. It’s about who embeds AI into the fabric of their economy the fastest and sets the rules everyone else has to play by.
Microsoft President Brad Smith said it best:
“The No. 1 factor that will define whether the U.S. or China wins this race is whose technology is most broadly adopted in the rest of the world. Whoever gets there first will be difficult to supplant.”
China called its shot in 2017: “Advocate open-source sharing and joint innovation.” That’s the strategy behind DeepSeek, GLM-4.5, and dozens more. Make it open, make it cheap, make it stick. Once you build on their stack, you don’t leave.
▼
💡How To AI: Automate Your Work With Gumloop
Gumloop made headlines with an ambitious goal: hit a $1B valuation with fewer than 10 employees. The platform lets users build AI-powered automations through a simple drag-and-drop interface, making it easy to connect apps, process data, and integrate AI without writing a single line of code. And yes, there’s a free version.
Here’s how it works:
Start with a Trigger - Pick an event that kicks things off, like a new row in Google Sheets or a form submission.
Add Actions - Define what happens next—send an email, update Airtable, or post to Slack.
Bring in AI - Drop in AI nodes to summarize documents, generate text, extract data, or perform web research.
Connect Your Apps - Integrate tools like Google Sheets, Gmail, Notion, or Airtable to move data automatically.
Customize with Logic - Add conditions and loops for smarter, more flexible workflows, no coding required.
Gumloop hints at a bigger shift: the future of work won’t belong to those who can code, but to those who can design systems. By turning AI-powered automation into something as intuitive as drag-and-drop, it lowers the barrier from “expert-only” to “anyone with an idea.”
▼
In Partnership With ClickUp
Meet ClickUp, the all-in-one AI-powered productivity platform that brings your tasks, docs, and teams together.
The Only AI That Knows All Your Work
Most AI tools start from scratch every time. ClickUp Brain already knows the answers.
It has full context of all your work—docs, tasks, chats, files, and more. No uploading. No explaining. No repetitive prompting.
ClickUp Brain creates tasks for your projects, writes updates in your voice, and answers questions with your team's institutional knowledge built in.
It's not just another AI tool. It's the first AI that actually understands your workflow because it lives where your work happens.
Join 150,000+ teams and save 1 day per week.
▼
Heard in the Server Room
Tesla just inked a $16.5B deal with Samsung to crank out its next-gen AI6 chips at Samsung’s long-delayed Texas plant, with Elon Musk saying he’ll “walk the line” himself to make sure things are up to scratch. For Tesla, it’s a bet on future self-driving cars and Optimus robots; for Samsung, it’s a lifeline for its struggling foundry biz and a shot at closing the gap to rivals. The chips won’t roll out for a few years, but the deal is a win for both companies struggling to turn their businesses around.
Researchers in China claim to have built the first AI system capable of autonomously designing neural architectures, reporting 100+ state-of-the-art discoveries without human input. If real, this would be a game-changer, eliminating human bottlenecks and accelerating AI innovation. But the paper’s clickbait title is a major red flag, and its tone feels more like ChatGPT than peer-reviewed science. We advise waiting for independent replication before bracing for an AI research revolution.
Meta just scored another big win in the AI talent wars, proving that money still talks. Shengjia Zhao, ChatGPT co-creator and former OpenAI lead scientist, is now chief scientist of Meta’s Superintelligence Labs, setting research strategy directly with Mark Zuckerberg and Alexandr Wang. The hire raised questions about Yann LeCun, Meta’s Chief AI Scientist for Facebook AI Research. His role remains unchanged, as he continues focusing on long-term research such as “world models,” which aim to replace today’s large language models. New hire Zhao will lead work on near-term advanced systems for “personal superintelligence.”
▼
In Partnership With Guidde
Turn your how-to’s and process flows into beautiful step-by-step video guides automatically with Guidde.
Create How-to Videos in Seconds with AI
Stop wasting time on repetitive explanations. Guidde’s AI creates stunning video guides in seconds—11x faster.
Turn boring docs into visual masterpieces
Save hours with AI-powered automation
Share or embed your guide anywhere
How it works: Click capture on the browser extension, and Guidde auto-generates step-by-step video guides with visuals, voiceover, and a call to action.
▼
🖋️ Anthropic Agrees To Sign EU Code Of Practice Despite Messy Past
Anthropic is signing the EU’s General-Purpose AI Code of Practice, reinforcing its transparency-and-safety-first image. The company says the Code, alongside the EU AI Act, strikes the right balance between innovation and risk management while aligning with its Responsible Scaling Policy.
Why sign when Meta refused? Anthropic likely sees value in cooperating with regulators. It builds trust, signals leadership in responsible AI, and could make future compliance easier. Meta, on the other hand, may worry about adding friction to its fast-moving product roadmap.
Still, Anthropic’s track record isn’t exactly spotless.
Pirated Books Lawsuit - A judge ruled Anthropic illegally downloaded over 7 million pirated books to train Claude, allowing a class-action lawsuit to move forward.
Reddit Scraping Allegations - Reddit sued Anthropic for allegedly scraping millions of user posts without permission to train Claude.
Book Destruction Controversy - Anthropic destroyed millions of physical books to feed Claude’s dataset.
Bottom line: Anthropic is leaning hard into its “responsible AI” branding, but its history shows why that narrative deserves scrutiny, especially since playing nice with regulators could turn out to be a very profitable move.
▼
That’s it for today! Have a great week, and we’ll catch you next time with more neural nuggets.