Sebastian Rodriguez

Entrepreneur & Scientist - #Innovation and #Sustainable #Energy

[tetsuo @tetsuoai]

Follow Tom Yeh. Easily one of the best ML educators on this platform.

[unusual_whales @unusual_whales]

Trump says they will be hitting Iran very soon.

Of note, US CENTCOM has struck Iran for nine consecutive days.

[Chubby♨️ @kimmonismus]

Kimi fixed all 15 critical bugs in 10 hours in a single prompt that GPT-5.6 and Fable 5 refused to fix because of guardrails.'
Respect for Chinese open source models is growing daily.

[The Transcript @TheTranscript_]

Databricks CEO: "We host open-source models such as Kimi and offer them to our customers. Demand has been so strong that we are running out of GPUs across multiple regions. We nearly exhausted our GPU capacity in Asia, and demand is rising in countries including Japan, South Korea, the United States, and India. We, therefore, need to acquire a large number of additional GPUs, which requires significant funding. That demand was what triggered our latest fundraising round: we were inundated with customer requests and needed more GPU capacity. GPUs are extremely expensive to acquire.

[huaijiang @huaijiangzhu]

my spicy theory is that chinese ai labs keep winning because of culture, not talent or resources.

chinese labs still have a deeply hands-on engineering culture. deepseek and kimi are flat organizations where science and engineering are fused: the same people move between algorithms, data, and infra, doing whatever it takes to make the model work.

but sf tech bros have decided that “researcher” is the high-status title while infra is merely support work. every new sf ai startup calls itself a “lab,” every ambitious engineer quietly upgrades their title to “research engineer,” and the infra work is left to whoever failed to escape it.

but at frontier scale, infra determines experiment velocity, and experiment velocity determines research output. at frontier scale, infra **is** research.

[Sebastian Rodriguez @sebrdz]

@RnaudBertrand @RnaudBertrand this is true - in addition China will sell the batteries and power electronics needed everywhere in the world to power AI

[Arnaud Bertrand @RnaudBertrand]

It's becoming clearer and clearer that China's AI open source strategy may end up being seen as one of the greatest strategic masterstrokes of all time.

They started with a clear resource and technological disadvantage - mainly due to the US semiconductor export controls - and have managed to change the rules of engagement in such a way that the US's own tech leaders and officials are now publicly siding with China's approach against their own companies. Which is pretty extraordinary.

When you can't fight symmetrically, make the adversary's way of fighting obsolete and self-defeating.

And the greatest irony in all this is - had the US not done the export controls - there's a decent chance that not only China wouldn't have gone for the open-source approach but the US would have made an enormous amount of money selling compute to them.

Now they're getting neither the money nor the containment.

[tphuang @tphuang]

WAIC 2026 is site of many SuperPoDs. HW unveiled Atlas-950 for the 1st time. Unlike its presentation last yr, this initial sales version only has 1024 NPUs & 1 EFLOPS compute + 256 TB memory & 3μs RTT delay.

Some photos of Atlas-950 network & compute cabinets. More to follow:

[Kautuk | Conscious Engines @Kautukkundan]

"India cannot build Frontier Language Models"

Bangalore Paper Club edition 2 happened Last night - The theme was alternate architectures for language models. Why?

Scaling is a compute game. Architecture is an ideas game.

One needs GPUs you have to buy. The other needs questions worth asking. If the frontier gets redrawn, it won't be by stacking more layers - it'll be by rethinking the ones we have.

4 papers were presented -

1. LLaDA - Large Language Diffusion Model
2. TwoTower - new architecture for Diffusion Language Models
3. CLEGR - new benchmark for Graph-Language Models
4. Dognosis - cancer detection via Canine Olfaction

Thanks for showing up and asking questions - My personal takeaway was that there are more people working in diffusion language models domain than I originally thought. And that's the whole point of running a paper club.

If you are one of them, let's connect.

[Sebastian Rodriguez @sebrdz]

@omooretweets Yeah - it is good to remember

[David Sacks @DavidSacks]

This is concerning. For the first time, a Chinese model Kimi K3 has taken #1 on the Frontend Code Arena and is scoring at or near the frontier on other benchmarks.

Meanwhile America is tying itself in knots: politicians and bureaucrats are banning new data centers, piling on state regulations, and pushing for new federal agencies to pre-approve frontier models.

This is how you lose the AI race. The rest of the world won’t play by our rules if we bog ourselves down. Permissionless innovation is how America won the internet and became the technological envy of the world. We can do it again with AI -- while addressing risks in a targeted way -- or we’ll watch our lead evaporate.

[Sriram Krishnan @sriramkri]

Such a fantastic read.

What a great quote.

"Every invoice in this industry is a confession about what the models still cannot do"

[Zephyr @zephyr_z9]

After the K3 drop, Huawei has now released a real 950 SuperPoD
1 EFLOPS fp8 & 2 EFLOPS fp4
256TB of unified memory

[Kyle Chan @kyleichan]

US-based AI company Thinking Machines says their new model was built in part on DeepSeek’s mixture of experts architecture and distillation on outputs from Moonshot’s Kimi-2.5.

[Sebastian Rodriguez @sebrdz]

@Xianbao_QIAN It would be great to have this. I am interested in how opportunities apply to markets other than US and Europe. I see demand in Mexico and Kenya for example.

[𝘊𝘰𝘳𝘳𝘪𝘯𝘦 @OopsGuess]

Trevor Noah and his guest discussed the serious disconnect in the Western world’s understanding of China.

One point was especially absurd:
many senior U.S. lawmakers and policymakers had spent their entire careers writing, voting on, and shaping China policy — yet only recently visited China for the first time.

They saw modern China with their own eyes and came back stunned, saying they had never imagined China had become like this.

This is simply unbelievable.

Ordinary Americans posting nonsense about China online is one thing.
But officials responsible for China policy being this ignorant about China is another level of imperial arrogance.

For decades, America froze China inside old stereotypes:
Bruce Lee.
Jackie Chan.
Kung fu movies.
Cheap toys.
Fake goods.
The world’s low-end factory.

Then China changed.
And they did not update the file.

China built high-speed rail, EVs, ports, drones, shipyards, AI models, supply chains, and the world’s largest industrial system while Western elites were still staring at a VHS tape from the 1990s.

But that is still not the real problem.

America’s misreading of China does not come from a lack of information.

It comes from American exceptionalism.

They grew up believing the U.S. is the natural center of democracy, freedom, technology, culture, money, and military power — the country everyone should admire, imitate, and obey.

So when China rises without asking for American permission, they do not see reality.
They see an error.

They keep saying Chinese people live behind a firewall.
But Chinese people translate, repost, study, mock, analyze, and consume more foreign information than most Americans ever bother to read about China.

Millions of Chinese study abroad.
Tens of millions travel overseas.
Chinese people use VPNs because they want to see the outside world.

Meanwhile, how many Americans have actually lived in China before confidently explaining China to the world?
How many can read Chinese platforms?
How many follow Chinese debates?
How many know what ordinary Chinese people actually argue about?

A lot of Chinese people criticize America because they have seen America.

A lot of Americans criticize China because they have seen headlines about China.

That is the difference.

China’s so-called “closed society” produced people curious enough to climb over walls.

America’s “open society” produced people too arrogant to look outside their own mirror.

[Thinking Machines @thinkymachines]

Today, we are introducing Inkling.

Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.

https://t.co/Ghebq5mG30

Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵

[Sebastian Rodriguez @sebrdz]

Good job

[Thinking Machines @thinkymachines]

Today, we are introducing Inkling.

Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.

https://t.co/Ghebq5mG30

Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵

[Zhengyao Jiang @zhengyaojiang]

More in the blog post:
- a breakdown of the discovered algorithms
- the rejected ideas AIDE² tried, covering a surprising share of the search literature
- the dead code it shipped
https://t.co/gbOZ1qlEfY
(7/7)

[Zhengyao Jiang @zhengyaojiang]

On our RSI ladder, AIDE² is Level 1.
Its self-improvement efficiency went beyond manual R&D with general AI tools, on held-out benchmarks.

We also tested Level 2, whether the improved inner agent makes a better outer loop. Results are mixed, and we do not claim ignition. (6/7)