We bet on tech founders, report and invest, then bring China's AI insider stories to the world. First angel in @UnitreeRobotics. Run by our reporter in SF.
Put two interviews with MiniMax (@MiniMax_AI)'s Yan Junjie (IO) side by side. You will get a pretty good snapshot of how AI has changed in 2 years.
In 2024, he told GeekPark (@GeekParkHQ) that pushing the frontier was the hard part. Cutting costs by 10x? There was already a playbook for that. He even predicted a nearly 100x drop in two years.
Fast forward to his recent podcast with Lawrence Burns from Scottish Mortgage (@ScotMortgage_UK). Now he calls low inference costs MiniMax's moat.
Why? Because cheaper inference lets you scale RL and push the frontier further.
How the thing that used to come after a breakthrough is now part of how you get to the next one.
If you want to know how Chinese entrepreneurs think about AI & Tech, watch this panel. youtube.com/watch?v=qtk2By66…
CATL’s Robin Zeng and Baidu’s Robin Li opened the APEC CEO Summit in Shenzhen this week, on the same stage as Rio Tinto CEO Simon Trott and US Chamber of Commerce president Suzanne Clark.
Clips worth highlighting:
2:47 Zeng on CATL’s two bets: battery swaps that take 99 seconds, and cells pushed to 350 Wh/kg so flying taxis can cover a two-hour drive between Asian cities in 15 minutes.
8:15 Li splits AI into four layers: chips, cloud, models, apps. He says value only shows up in the last one. His example is robotaxis, which he puts at roughly 10x safer than human drivers, and he wants more Asia-Pacific economies to open their roads.
10:41 Li’s next bets after robotaxis: AI drug discovery, and AI-made short dramas. “I do care about application more than the models or graphic chips.”
11:17 Zeng turns interviewer and asks Li whether today’s models are enough for science. “We try to find AI Einstein.” Li’s answer: AI can grind on a defined problem for months, but it still can’t ask a good question.
27:02 Li on regulation: too much slows innovation, too little brings risk, and the US and China face the same dilemma. Then compute: even US frontier labs are short. Baidu just launched a token exchange that matches “token factories” with companies that need inference.
30:01 Zeng on how rules become culture, using drunk-driving laws as his example.
35:52 Li tells his own staff to “do less, not more.” Focus forces you onto partners, which is why Baidu tests robotaxis with Uber and Lyft in London and is helping Dubai’s transport authority write its autonomous-driving rules.
The whole conversation runs 39 minutes.
Everyone is thinking about what robots might one day do with their hands. Few are thinking about what it takes to build those hands.
At the Tokens& (@tokensandai) Frontier Models panel, Ted Xiao (@xiao_ted), now at Bezos-backed Prometheus (@PrometheusInc) noted that roughly 75% of physical AI investment goes into dexterity models.
But intelligence still needs chips, factories, and raw materials.
“When you expand the scope, a lot of the fruit is very low-hanging.”
Perhaps the most overlooked part of building intelligent machines is everything that must exist before they can move.
Why NVIDIA wouldn’t be NVIDIA without Jensen Huang. Five moments from NVIDIA’s history are worth remembering.
1. Admitting the technology was wrong.
In the 1990s, Jensen went to Japan to tell Sega that NVIDIA couldn’t deliver the technology it had promised. He recommended that Sega find another supplier. Then he asked for the money NVIDIA still needed to survive.
Sega provided $5 million.
Before he could ask someone to believe in NVIDIA’s future, he had to tell the truth about its present.
2. Paying for something the market didn’t yet value.
In 2007, NVIDIA began putting CUDA into all its GPUs. Gaming customers had little reason to pay for general-purpose computing. Shareholders questioned the expense. The returns took years.
Jensen kept investing.
What eventually became a competitive advantage first appeared on the accounts as a cost someone had to keep defending.
3. Leaving a future he had once believed in.
NVIDIA entered the smartphone-chip market in 2010, then withdrew as it commoditized. Years of effort could have become a reason to stay. Jensen accepted the retreat, opening room for robotics and automotive computing.
Conviction had helped him begin. Changing his mind allowed the company to continue.
4. Giving employees greater certainty when he couldn’t give them answers.
At the start of COVID, NVIDIA brought raises forward by six months and continued paying contractors.
Nobody knew how the pandemic would unfold. But the company could decide when money reached a household. It could make one part of an uncertain life more dependable.
5. Leaving room for a person to become more capable.
Asked about his reluctance to fire people, Jensen said: “I’d rather improve you than give up on you.”
He remembered having cleaned bathrooms himself. His own beginnings gave him a reason to hesitate before treating someone’s current limitations as permanent.
None of these moments could guarantee the NVIDIA we know today. Together, they reveal how much of a company’s future depends on decisions made before that future can be justified.
A balance sheet records the money spent. It cannot fully record the patience behind it, the embarrassment endured to obtain it, or the people given enough time to become worthy of it.
Entrepreneurial spirit is the soul of a company. It takes shape in what a leader chooses to protect, what they are willing to reconsider, and what they continue to believe is possible for other people.
No Jensen, No NVIDIA. @JensenHuang@nvidia
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me.
I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA.
I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation.
Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO.
Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours.
Jensen
Jensen Huang has made his first statement since resigning, and despite media reports to the contrary says that he would be honored to answer the call to return as NVIDIA CEO:
'Since my departure, I've seen an outpouring of calls from Nvidia's employees, investors, business partners, and customers for me to return as CEO. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO'
In late October, GeekPark spoke with Pine AI (@pine_computer) CTO Vincent Sun (@vincent_sun_bro), who gave us an early look at Pine Phone three weeks before its launch.
Despite the name, Pine Phone isn't hardware. Neither is Pine Computer. Both are software APIs designed to let developers build AI agents that can interact with the real world.
A few things that stood out from our conversation:
1. Why AI needs its own computer
"This world wasn't built for AI. It was built for humans."
Sun argues that today's models are already capable, but the interfaces they rely on were designed for people.
A turning point came when a user needed to cancel a flight after being summoned for jury duty. The airline's phone agent couldn't resolve it, and its website offered no way to handle the request.
"Phone calls alone couldn't solve it, so we started building a computer."
Pine went on to redesign the OS and browser around how models process visual information, focusing on what changes rather than repeatedly processing the entire screen.
Sun claims the approach allows models costing 20–25x less to complete tasks 2–5x faster, with better results.
2. Why Pine won't default to "always approve"
"We won't."
Sun draws a distinction between AI operating on a computer and AI acting in the real world, where mistakes can carry much greater consequences.
3. Phone vs. computer
"Voice is how you interact with real people. A computer is how you deal with the portals and interfaces people built."
Sun doesn't expect either to replace the other anytime soon.
4. Why make it available to developers?
"We're not fixated on any one product form."
Rather than keeping the technology exclusive to Pine's own products, Sun wants other developers to build on it.
The broader ambition, he said, is to give AI the ability to get things done in the real world.
This week, Pine Phone became available to Muse and Instinct users.
Congrats on the launch! @StanleyWei4748@vincent_sun_bro
A month ago, we launched Pine Computer—an AI-native computer for agents.
What developers built next surprised us.
Today, we're open-sourcing Pine Drama Studio and bringing Pine Phone to Pine Computer.
Here's the bigger story. 🧵
ESTOS 39 MINUTOS DEL FUNDADOR DE KIMI SON ORO PURO
explica cómo montó desde China una empresa de IA valorada en más de 50.000 millones
hay másteres de 20.000 € que no enseñan lo que él cuenta en 10 minutos.
39 minutos, GRATIS y sin relleno.
guárdalo y mira esto cuando tengas tiempo, me lo agradecerás
Perplexity (@perplexity_ai) CEO Aravind Srinivas (@AravSrinivas) spent 15 minutes at SGLang Summit (@sgl_project) talking about “Building the Next Computer.”
Some notes from the talk:
00:44 - How it started. Perplexity built a Slack agent in January where employees talked to a bot more than each other. Marc Andreessen texted Aravind about a company where employees mostly talked to a bot instead of each other. “That was basically us.”
01:35 - Why Computer? Almost SuperAgent or MasterAgent, but he chose Computer, a word that once referred to people doing math for astronomers.
04:20 - Search as code. Fetch, chunk, embed, retrieve, all exposed through a Python SDK so agents can build their own search pipelines.
08:15 - SPACE. Perplexity now runs millions of concurrent sandboxes every day. The platform is open source, including fixes from red-teaming.
12:22 - Local + cloud. A fine-tuned Qwen 27B scored 59.6% locally on 89 tasks; calling Opus 5 only when needed raised it to 73%.
13:20 - Lily. Built by an intern who joined full-time, this Apple Silicon inference engine beats MLX-LM by ~22% on prefill and ~30% on decode on an M5 Max.
He ended by telling everyone they're still hiring.
Jensen Huang matters to Nvidia because he can take technologies his engineers believe are possible and turn them into products the whole company commits resources to, takes risks on, and has to deliver. That's why the supply chain gets such strong order visibility and is willing to keep up with aggressive spec upgrades. It also forces competitors to follow, which drives technology upgrades and supply chain investment across the entire AI industry.
Between the two, deciding on aggressive spec upgrades and making them happen is the harder part, compared with giving the supply chain order visibility. Huang doesn't need to be the person at Nvidia who knows the most about new technologies like HVDC, CPO/NPO, or CCL/PTFE. His value is that he weighs strategy, organization, technology, and supply chain all at once, then makes the call quickly and puts resources behind it.
Based on what's public so far, it looks hard for Huang to come out of this unscathed. If so, the practical move may be to gradually reduce Nvidia's reliance on Huang's personal decision-making while preparing a succession plan. Along the way, two things are worth watching to gauge where AI technology and investment are heading: whether Nvidia can keep developing new products and technologies as efficiently as before, and how competitors respond. Supply chain research will be an important basis for that judgment.
Here's the prompt we gave Claude:
"Build a first-person shooter game. Players should be able to choose their character and weapons before entering an open map where they can run around and shoot.
Use WASD for movement, mouse movement to look around, and left-click to fire.
The graphics don't need to be fancy, but the gameplay has to feel smooth and responsive."
That's it. The Future is coming 🫥