FounderCoHo

Share founder story.
Fuel founder journey.

Substack: | Youtube:

[Jing Conan Wang @jingconan]

AI coding is letting us ship software 10x to 50x faster.

But here's the cold, hard truth: it is not making your business run any faster.

Why? Let's break down the dangerous illusion of speed in the AI era. 🧵

[FounderCoHo @FounderCoHo]

Agentic GTM: How a16z, Khosla & HF0 Builders Automate Customer Acquisition. 👇

Distribution is now the only moat that matters, as @evanspiegel recently put it. AI makes products easier to build and copy, so the real advantage is how fast you reach, retain, and scale your audience. This is what we’ll breakdown on our Upcoming Event

We’re bringing together founders and operators building the next generation of AI-native growth:
• Lucky Zhang (@lucky_z2), Founder & CEO of @sekaiapp
• Seijin Jung (@SeijinJung), Co-Founder & CEO of @enrich_labs
• Bolbi Liu (@BolbiLiu), Founder & CEO of @AdsGency_AI
• Sandra Li (), Head of US Enterprise at Antom by Ant International

We’ll cover how AI agents are reshaping growth across mini-app engagement, ads, content, SEO, 24/7 campaign execution, and leaner paths from zero to scale.

You’ll be in the room with founders who’ve scaled to millions and leave with playbooks you can use right away.
Register: https://t.co/7HA15FOva8

[Jing Conan Wang @jingconan]

Most startup founders are building the wrong moat.

They focus on:
- Technology: easily replicated.
- Distribution: easily bought.

But they completely overlook the most powerful defensibility there is.

I call it the Organizational Moat. Let's break it down 🧵

[Jing Conan Wang @jingconan]

AI gave solo builders the power of an entire 10-person team. But it also created a dangerous trap: the "Digital Solo Builder" echo chamber.

If you are building an AI startup from a laptop, read this. Here is the solo builder's silent killer: 🧵

[FounderCoHo @FounderCoHo]

Hello everyone! We've compiled a list of upcoming events and exclusive resources for the FounderCoHo community. Feel free to check it out and participate!

👉 https://t.co/TOCtH0kqAu

[FounderCoHo @FounderCoHo]

Why Anna left Google to start a company:

"At a large company, you're playing a game of chess but you can't move the pieces."

[FounderCoHo @FounderCoHo]

$4B valuation. $300M Series A. The bet?
AI that designs its own chips. 👇

Anna Goldie (@annadgoldie), Founder & CEO of @RicursiveAI and co-creator of AlphaChip, is building toward a recursive loop: AI designs better chips → better chips train better AI.

Today, frontier chips can take years and hundreds of millions to build.

Ricursive’s Vision :
▸ Speed up chip design,
▸Make custom silicon more accessible,
▸Eventually co-evolve AI with the hardware that powers it.

AI is no longer just running on chips.
It’s starting to redesign them.

[FounderCoHo @FounderCoHo]

LLMs recreate us. Specialized AI does what we never could.

If it's something humans are good at - language, code, intent, use an LLM.
It does what we do, just faster and at scale.

If it's something humans can't do, like placing billions of components under thousands of physical constraints, you need specialized AI.

[FounderCoHo @FounderCoHo]

Full conversation with Anna Goldie
👉 https://t.co/ZPtZsFYzAk

[FounderCoHo @FounderCoHo]

A 3-phase plan to remake chip design:

Step1: Fix the worst bottlenecks (physical design, verification) so Nvidia, AMD, Samsung ship faster

Step2: Open it up as a platform so anyone with real workloads can design custom silicon

Step3: Design our own chips, train our own models on them, and let the two co-evolve

Recursive self-improvement, in hardware.

[FounderCoHo @FounderCoHo]

$4B valuation. $300M Series A. The bet?
AI that designs its own chips. 👇

Anna Goldie (@annadgoldie), Founder & CEO of @RicursiveAI and co-creator of AlphaChip, is building toward a recursive loop: AI designs better chips → better chips train better AI.

Today, frontier chips can take years and hundreds of millions to build.

Ricursive’s Vision :
▸ Speed up chip design,
▸Make custom silicon more accessible,
▸Eventually co-evolve AI with the hardware that powers it.

AI is no longer just running on chips.
It’s starting to redesign them.

[FounderCoHo @FounderCoHo]

The mentality every successful founder has:

"This company needs to exist and it has to be me who builds it"

[FounderCoHo @FounderCoHo]

$4B valuation. $300M Series A. The bet?
AI that designs its own chips. 👇

Anna Goldie (@annadgoldie), Founder & CEO of @RicursiveAI and co-creator of AlphaChip, is building toward a recursive loop: AI designs better chips → better chips train better AI.

Today, frontier chips can take years and hundreds of millions to build.

Ricursive’s Vision :
▸ Speed up chip design,
▸Make custom silicon more accessible,
▸Eventually co-evolve AI with the hardware that powers it.

AI is no longer just running on chips.
It’s starting to redesign them.

[FounderCoHo @FounderCoHo]

Full conversation with Anna Goldie
👉 https://t.co/ZPtZsFYzAk

[FounderCoHo @FounderCoHo]

$4B valuation. $300M Series A. The bet?
AI that designs its own chips. 👇

Anna Goldie (@annadgoldie), Founder & CEO of @RicursiveAI and co-creator of AlphaChip, is building toward a recursive loop: AI designs better chips → better chips train better AI.

Today, frontier chips can take years and hundreds of millions to build.

Ricursive’s Vision :
▸ Speed up chip design,
▸Make custom silicon more accessible,
▸Eventually co-evolve AI with the hardware that powers it.

AI is no longer just running on chips.
It’s starting to redesign them.

[FounderCoHo @FounderCoHo]

We've compiled a list of upcoming events and exclusive resources for the FounderCoHo community. Feel free to check it out and participate! 👇
Link: https://t.co/TOCtH0kqAu

[FounderCoHo @FounderCoHo]

Agreed, and the underrated part runs deeper than ownership.

Liang stockpiled 10,000 A100s inside the hedge fund before the AI race started, resisted outside capital for nearly 3 years, and hasn't spoken publicly about R1 once.

His V4 motto, from xunzi:
"Not seduced by praise. Not cowed by slander. Follow what is right. Hold yourself straight."

The full story of how he got here 👇 https://t.co/aWzLFJT79D

[FounderCoHo @FounderCoHo]

Love this story. It tracks with everything else about him.

A decade quietly stockpiling 10,000 A100s out of curiosity. Three years saying no to outside capital. Silent through R1's global moment.

Liang isn't optimizing for the obvious win. He never has.
the long version 👇

https://t.co/aWzLFJT79D

[Abhinav Kukreja @kukreja_abhinav]

My favorite liang wenfeng story is that when he was in college, a buddy offered him a cofounder position in his drone startup. He turned him down to focus on academics. The drone startup became
DJI.

You would assume making such a decision would be the defining career moment of someone’s career. That they would regret it forever. But not this guy 😅

[FounderCoHo @FounderCoHo]

Before reading too much into V4.1, it helps to know who's behind it.

Liang wenfeng hasn't spoken publicly about R1 once.

He ran a top Chinese hedge fund, stockpiled 10,000 A100s before the ban — out of curiosity, not strategy, and resisted VC for 3 years.

The most mysterious lab in AI, unpacked 👇 https://t.co/aWzLFJT79D

[Jukan @COMPUTEX @jukan05]

The Information: DeepSeek is expected to release V4.1 this June.

The Information: DeepSeek is working to turn its technology into products and services it can charge customers for. As part of these efforts, DeepSeek has been hiring talent from ByteDance’s product development teams and other Chinese tech companies.

The Information: DeepSeek’s new V4.1 model is expected to offer more tools for enterprise users and better support for MCP, or Model Context Protocol, a standard for connecting AI models with other software. According to people familiar with the matter, DeepSeek plans to make V4.1 capable of processing images and audio, although it will still only be able to generate text as output.

[FounderCoHo @FounderCoHo]

All true. the wilder part isn't in here:
1) Liang resisted outside capital for ~3 years. This $7B isn't his latest round. It's his FIRST.
2) Lead investor: China's state-backed "big fund," not a VC
3) 97% of the core V4 team stayed through brutal poaching
4) DeepSeek denied nvidia early access to V4 and is migrating inference to Huawei
5) Liang Wenfeng is not running the OAI/Anthropic playbook. He's building a sovereign AI stack.

Read the philosophy behind it 👇
https://t.co/aWzLFJT79D

[Ejaaz @cryptopunk7213]

deepseek is raising a monster $7 billion round at $50B val making it china's largest ever AI raise but what shocks me the most is the founder, liang wenfeng:

> he's personally contributing 40% of the round himself. $3 billion.

> he owns 90 PERCENT of the company (unheard of at this valuation)

> deepseek was founded inside his hedge fund, one of China's most successful funds.

guys a fucking beast. this round is meant to achieve 2 things:

1. acquire as much compute to push out new deepseek models more often

2. turn deepseek revenue-positive by pushing new enterprise products (same tactic as OAI and anthropic)

deepseek v4.1 is expected to release soon.

[FounderCoHo @FounderCoHo]

Every AI lab is fighting for attention.
DeepSeek is the striking opposite.
They ship. Then they go silent. Liang Wenfeng hasn't spoken publicly about R1 once.
When V4 dropped, they shared a single line from Xunzi:
"Not seduced by praise. Not cowed by slander. Follow what is right. Hold yourself straight."
Then slipped backstage again.

18 years from a secretive hedge fund to an AGI lab — built on three counterintuitive bets:
1. A 10-year infra buildout driven by curiosity, not strategy (10,000 A100s before the ban)
2. No VC for almost 3 years
3. Embracing alpha decay by giving the recipe away

Read the full story here:
https://t.co/aWzLFJT79D

[FounderCoHo @FounderCoHo]

@greg16676935420 Tomorrow in Beijing:

[FounderCoHo @FounderCoHo]

When OpenAI released ChatGPT, the team was worried about hallucinations.

But it reached 1 million users in just 5 days.

People were so hungry for it that they didn't care it's imperfect.

The same is happening in the medical field.

"A very common chore we do is to scan through a patient's medical record. It can be hundreds of thousands of pages long."

To make this process less painful, Stanford Professor Jonathan Chen and his graduate students built chatEHR.

It's still in beta, but people are already desperate for it.

[FounderCoHo @FounderCoHo]

Doctor + AI can be worse than AI alone. Why?

That’s what Stanford Professor Jonathan Chen (@jonc101x) discovered in a landmark GPT-4 medical study. Doctors using AI performed about the same as doctors using Google — while the AI alone outperformed both.

The bottleneck isn’t the model.
It’s human-AI integration.

We talked about:
• Why “human + AI = best results” may no longer be true
• How AI reinforces diagnostic anchoring bias
• Why many doctors still use AI like a search engine
• Why innovation happens at the intersection of disciplines

“Safety isn't what slows down innovation. Safety is what enables you to innovate boldly.”

[FounderCoHo @FounderCoHo]

Stanford professor & doctor: "Human in the loop is disappearing in many medical tasks."

So what does that leave for doctors? 👇