lukex

co-founder, chief janitor | venture partner

[lukex @Lukex]

@TangTrades Luke obvi

[lukex @Lukex]

@CQ_Neverland @Waymo in 2 years anyway 😂

[Paige ✈️ CoRL @paigeinsf]

Second mover advantage is underrated.

[Artem @sl1ma4]

POV: you assigned 4 Gemini agents to fix a single bug.

[Bunagaya @Bunagayafrost]

USG "Hey China, can you submit your open models for testing before release?"

China "Oh, go right ahead. It's available online for anyone to test"

USG "...No I mean like submit it to us before releasing it online"

China "Like I said, it's available online. Do you not know how to download open models?

USG "Of course we do, but before publicly making it available for everyone we want you to submit it to us exclusively for safety testing"

China "Oh, oh I see. No."

USG "...If you don't cooperate we may have to restrict access to your models"

China "How?"

USG "By forbidding our enterprises from using them"

China "OK, force them to pay for expensive tokens. I don't care. Nice talking to you"

---

Staff "Sir China won't cooperate"

President "Xi is a tough cookie"

Staff "Sir, if we don't safety test Chinese open models, but safety test the domestic open source, we will be giving China an unfair advantage"

President "Fine, just safety test the closed ones. They're the ones begging for it"

[lukex @Lukex]

@thsottiaux + actively suggesting monitoring is huge

[lukex @Lukex]

@thsottiaux Convo forking has been a big differentiating feature + much more generous limits!

[Jeff Wang @jeffwang]

Exactly one year ago was the craziest 72 hours I’d experience in Silicon Valley. 

I was thrust into a situation where many members of the Windsurf team had moved on to Google, recruiters were reaching out to every member of the team, customers were calling to understand what was happening. It was definitely the scariest time of my life, and I was caught in a blur of nonstop phone calls and meetings to figure out what to do next

But time seemed to slow down when me and Graham met with Scott and Russell. Everything seemed to click and make sense. They were looking for a GTM org, and we were looking for an Engineering org, both of which had been considered the best in Silicon Valley. Our products also did not overlap, we had been trying to build a cloud agent, they had been trying to build an IDE. It almost seemed too perfect, and 72 hours later we didn’t sleep to get the acquisition done. 

Now the part that isn’t told as often is what happened next. Windsurf quickly released Wave 11 and then a slew of other new features. We had a massive bug bashing effort to really polish the experience. Over the last year this led to unifying the platform around Devin.

Now the future is ahead of us to continue to shape. Windsurf was the first agentic IDE, Devin was the first cloud agent. We’re now seeing new form factors quickly emerge with our customers. For example event driven agents are agents that automatically take on tasks based on triggers, conversations, and general awareness about the company. Devin can also manage swarms of agents, each with their own computer to get work done, as an example our security swarm is one of the most popular new features across our enterprises. What happened a year ago didn’t slow us down, it’s made us go faster, and we’re looking forward to building the future of software engineering.

[Bill Gurley @bgurley]

WeChat has been a little quiet on AI front & perhaps even falling into being considered a laggard. But this week, Pony Ma presented a new vision on how to turn "mini-apps" (still no real equivalent in US) into AI handshakes. Manny partners announced. Could be a powerful move. Read more below.

[lukex @Lukex]

@ccatalini @openstandard Great analysis sir!

[Christian Catalini @ccatalini]

1/ Today, more than 140 companies, most of which compete fiercely with one another, agreed to back the same stablecoin. The vehicle is @openstandard, a new and deliberately independent company launching Open USD, or OUSD, and positioning it not as anyone’s product but as neutral infrastructure for payments, trading and the internet economy.

[Thierry from arvy 🇨🇭 @ThierryBorgeat]

🚨 SpaceX just pulled off the greatest financial engineering feat of the century. In about a week.

Here's everything that happened, in order:

– Folded xAI into a rocket company, turning "space logistics" into an "AI infrastructure" story overnight
– Priced the IPO at a flat $135. No book-building, no range. Take it or leave it
– Floated just 4% of the company. 556 million shares against 13 billion
– Raised $75 billion at a $1.77 trillion valuation, near 100x revenue
– Lobbied to get into major indices in ~15 trading days. Amazon took years. Forced buying, by law
– Handed an unusually large slice of the float to retail. Tiny supply, an army of buyers
– Watched the stock rocket past $200, up nearly 20% in a single session
– Saw ~46% of the entire float trade hands in one day
– Then announced a $60 billion all-stock buyout of Cursor, the AI coding tool
– Structured it so the higher the stock trades, the fewer shares it has to print to pay

A company losing $4 billion a quarter is now buying AI startups with paper it manufactured out of a 4% float.

The scarcity that pumped the stock now makes its shopping spree cheaper.

This isn't aerospace. It isn't even AI.

It's the finest financial engineering of the century, and it's only week one.

[lukex @Lukex]

@Cointelegraph Actual space stuff is a drop in the bucket compared to AI apps and infra 😂

[Andrej Karpathy @karpathy]

This is a super exciting release - Claude Fable 5 is the same underlying model as Mythos but with added safeguards. The benchmarks are great and it's SOTA on everything by a margin but I'll add that *qualitatively* also, this is a major-version-bump-deserving step change forward (imo of the same order as Claude 4.5 was in November), peaking especially for long problem-solving sessions on very difficult problems. You can give it a lot more ambitious tasks than what you're used to, the model "gets it" and it will just go, and it's never felt this tempting to stop looking at the code at all (but don't do this in prod!). The model still has quirks that people will run into and the safeguards are configured to be a little too trigger happy for launch, which can hopefully be tuned over time.

I feel a lot of things changing as working software increasingly comes out on a tap. The Jevon's paradox kicks in and I feel my own demand for software growing substantially. You can ask for anything - explainers, visualizers, dashboards, bespoke single-use apps (e.g. a full wandb that is hyper-specific just for your project), you can 10X your test suite, auto-optimize code, run giant research projects with custom HTML for the results, anything! "Free your mind" (Matrix ref). Really looking forward to all the things people build!

[Brian Armstrong @brian_armstrong]

Good take

My guess is
- demand for intelligence is near infinite
- but 80% of workloads will be running on 99% cheaper models within 12-18 months
- 20% of workloads will still run on latest gen models where IQ maxing is important (scientific breakthroughs, higher level ochestrator agents?)
- rough analogy might be what % of macbooks or gaming PCs sold have the maxed out specs for CPU/GPU, prices are falling much faster than Moore's law here though
- this leads me to think the limiting factor will be energy and compute, not better models

At Coinbase we're working hard on routing prompts to cheaper models where appropriate, and in some cases have been able to keep costs roughly flat, while token usage continues to grow exponentially.

[Spencer Yang @spenceryang]

SpaceX may soon become one of the first companies to IPO at a $2T valuation, bringing together SpaceX, xAI, and X.

I started my post-college career at Twitter. I watched the platform evolve, grow, struggle, reinvent itself, and even later worked out of its former San Francisco office after it became a co-working space run by BLK71 SF.

To mark the moment, I shipped TrillionMarketCap: a live registry of the assets, companies, commodities, and networks large enough to be measured in trillions.

Gold. NVIDIA. Apple. Bitcoin. SpaceX.

[Spencer Yang @spenceryang]

https://t.co/E44BsFddpZ

[Allie K. Miller @alliekmiller]

The most expensive mistake in enterprise AI right now: treating FDEs as your whole transformation plan.

Forward deployed engineers (FDEs) are important for custom deployments, but they won’t fix the change management issue most enterprises are facing.

It’s likely more the former that Anthropic and OpenAI will continue to prioritize (and hire into the thousands, who knows). Beyond performance and cost, it’s systems integration, ROI, and literal usefulness that drive revenue and stickiness.

*However*

External FDEs, in my opinion, will not make your company an AI-first company.

You can have the sleekest multi-agent orchestrations and still have the majority of your employee base hating AI, avoiding AI, and distrusting leadership decisions on AI.

And we already know this because we see this in traditional SaaS too: you can customize the heck out of your Salesforce deployment, but that doesn’t mean your sales team will improve their data hygiene or even attempt to change the way they track and grow with it.

Buying a fancier car doesn’t mean you magically learn to drive better overnight.

If you’re an enterprise exec and FDEs are sold as the immediate and sole solution to your company transformation woes, walk away.

It’s the combination of tech *and* people enablement *and* process reinvention that compounds into actual business outcomes.

Large complex enterprises will stall out if they only prioritize the first.

[lukex @Lukex]

@yoheinakajima consistently surprises and delights, the man IS the bleeding edge of AI use cases

[Yohei @yoheinakajima]

ran openai's 663 open roles through chatgpt to ask what one could infer about their strategy

nothing particularly new, but it worked well

[tae kim @firstadopter]

Oh look! Anthropic's entire "we are delaying Mythos" narrative was marketing hogwash.

Kudos to FT for confirming what was obvious. Anthropic simply doesn't have the compute.

FT: "Multiple people with knowledge of the matter suggested Anthropic was holding back from a wider release until it could reliably serve the model to customers."