Colin Campbell

The hyperscalers have already outspent the most famous US megaprojects

@ContrarianCurse yes but who gets the margin

All software is is a set of tools that are used repeatedly. They combine to create a process and output Especially in a compute constrained environment, which we will be probably for years, it is incredibly expensive and wasteful to recreate those tools over and over again. What has been baked into those tools is also years of domain expertise and network effects that the labs won’t be able to touch. Agents are just going to use your computer in the way people do. Now they might not need UI and yes for some the output can be achieved directly with AI - those suites are in trouble But for the most part, AI will go out and call on the same already built tools because that will be the cheapest way to get the job done

@BucknSF @MRatable @BucknSF bump on why or feel free to DM

@MRatable netskope a shitco

There was a time not too long ago when every hedge fund bro was ADAMANT that Netskope was going to kill Zscaler

had a dream i raised $73M from Sequoia to replant northern California’s costal redwoods.

@JesusFerna7026 @SamPassey3 x2 Chad jones the goat

@SamPassey3 Well, you are going to Stanford GSB. Chat with Chad Jones. He is the GOAT of this topic!

I concluded my Henry Family lecture at the University of Miami last Thursday by saying: “Two things are important right now in life: deep learning and fertility. Everything else is noise.” We are only starting to glimpse what these two forces will do to global life over the next fifty years. And they interact: deep learning will reshape demographics, and demographic collapse will reshape automation. Nearly all my posts on X (except some parochial commentary on Spanish economic policy) revolve around these two facts. So does most of my current research. Even work that does not seem directly connected turns out to be, once you look carefully. My papers on geoeconomics and international macro are about figuring out some of the consequences of deep learning and fertility. For example, my work on China focuses on its abysmal demographic future and how the U.S. is positioning itself (rightly or wrongly) to address it. And my work on political polarization and the welfare state is about the consequences of decades of low fertility in Western Europe. When people talk about political change in Western Europe, they are talking about low fertility, whether they know it or not. It is not clear that modern representative democracy can survive sustained fertility rates of 1.3. I do not say that with glee. The reason I decided to spend my life on academic work in economics is that I realized, when I was much younger, that daily events are irrelevant. The things that concern the media and 99 percent of commentary on X are largely irrelevant. One political party does better or worse in the next electoral cycle because of internal fights or a good campaign. At a fundamental level, none of it matters: the political outcome 25 years from now will not depend on those accidents. As Alexander Gerschenkron said, Clio is not a tidy housewife. The rise of any political movement is always full of advances and retreats. Social change waxes and wanes. But at the end of the day, as my favorite historian Fernand Braudel put it: “The events of history are merely surface disturbances, crests of foam that the tides of history carry on their strong backs.” or in the much better original: “Les événements de l’histoire ne sont que des agitations de surface, des crêtes d’écume que les marées de l’histoire portent sur leur dos puissant.” The tides of history today are deep learning and fertility.

@atelicinvest >youth hockey testing business what in the everloving alberta is this

This won't be news to many but since we are constantly bombarded with "oh look at this tradesperson they built their entire back end using x it's a revolution" Here's some sweet sweet anecdata. I have a smart buddy that's got a nice youth hockey testing business. His business is quite seasonal. Most of testing gets done in Sept and he's pretty free during rest of the year Being the enterprising dude he is, he has done exactly what all of these stories claim - he built out a full player evaluation management suite using Claude code. He's been laboring over this for 6+ months now. This isn't some toy shit. It's got a lot of nifty features that he needs. And it runs and deploys. It's quite impressive, and follows the story to a T. Knows exactly what he knows for his business. Knows what real world scenarios look like. He's was just in the process of putting it into production. Then really started to ask himself: "wait - do I *really* know this won't blow up on me?" So he called me up to take a look at what he's built. It's exactly what you might imagine - weird architectural choices, patches upon patches, some edge cases that will blow up the app in production. But - and here is the critical part - because the code base is so weird, there's no way for me to help him without fully understanding what he's done. And without knowing how the code works, it's hard to know what will break when we apply the fix, many of which will touch multiple areas of the codebase. This is not a terminal problem - the solution here is for him to level up his understanding of the tech so that he can build things properly - at least to a place where he can review the suggested changes and assess if it makes sense or not. I'm sure he will get there over the next few months. But I don't think many people have 8 months to 1 year of free time to do this. Can this get fixed with better models? To a degree - if what you need is super simple and within the model's capabilities then sure, he can just rely on the LLM. But most software isn't just frozen in time and space - when you do one thing you want to do 10 more things - complexity and scope creep are unavoidable in most areas.

@ContrarianCurse What is HALO

I dunno, I think the whole HALO trade is pretty well considered. Going asset light and finding characteristics of durability / conviction on terminal is likely a great alpha hunting ground currently

@CatNyanpital genie????? robotics????

This is exactly what I’m talking about Google enslaving Deepmind for search and killing off any other “unecessary” endeavors

Introducing FAUNA. The creative agent built for people whose ideas deserve better. Describe what you want to make. It builds the workflow on the canvas in front of you. Redirect it, push it further, tell it what to avoid. Your vision drives everything.

@BucknSF @MRatable why

Congrats to all my homies at Rhoda

We’re acquiring Promptfoo. We’re buying Safetysnoot. Proud to announce that the Ragschlorp team is joining us. We own a majority stake in Chunkwad.

Vail Resorts ($MTN) likely to open tomorrow down to where if you invested 10 years ago you’d have done as well putting your money in a hole. It’s time for a change to become more asset-light, sell off resorts, and allow character and differentiation to return to skiing.

“Venture capitalists (VCs) are increasingly abandoning their traditional role as monitors of their portfolio companies. They are giving startup founders more equity and control and promising not to replace them with outside executives. At the same time, startups are taking unprecedented risks—defying regulators, scaling in unsustainable ways, and racking up billion-dollar losses. These trends raise doubts about the dominant model of VC behavior, which claims that VCs actively monitor startups to reduce the risk of moral hazard and adverse selection.” Risk Seeking Governance Brian Broughman and Matthew Wansley

@PythiaR @k3ithmccullough this is as good as “a concept of a plan” or “a fact-based feeling”

@k3ithmccullough My book is the way it is because of my belief

Software was sold too hard, lot of semis are expensive But nothing has changed in terms of AI probably making software a substantially worse business than it was before The reversion is a trade, not a structural stance you should be taking IMO

@nabeelqu open.spotify.com/track/5rAff9Xf…

My LinkedIn is just the usual stream of company announcements and cringeposts, and then there's George Hotz poisoning the stream with these highly entertaining blackpills:

Pete Hegseth to Claude:

All my doctor friends be like “I’m doing a fellowship” constantly. Lord of the rings ass career.