Alex Heath

Alex Davis says the only thing he wrestles with before writing a check is regulatory, meaning whether a regulatory problem was a reason to pass. The price is not the wrestle. He published a letter saying value investing is dead, and he allows that the line may have been taken out of context. The claim underneath is that nobody knows where the valuation of most of these companies ends, or what the ceiling is.

Anthropic, two years ago, is the example. A model that marked the company at $30 billion and called the result a 2x would have missed a path to $100 billion of annual recurring revenue inside two years. The habit he is arguing with is buying a good company only at a discount. He states

Alex Davis expects new prime contractors for the first time in 100 years, and a defense-tech market that does not look like the old one. Working at Lockheed was not the job people wanted, and starting a company on that scale used to be too large a mountain. He thinks that company can be built now.

Anduril and Shield are the first two he sees starting to look like primes, and he expects more after them. Open models, in his view, can become the next large defense companies. How those groups end up working together is still open, and he calls the opportunity massive. He also says the model companies themselves are becoming defense companies, including work with the Pentagon, and that protectin

Alex Davis does not think there are too many neoclouds. He thinks there are not enough good ones. Alex Heath reads back a line Davis wrote: back owner-users, not speculative landlords, because this will not end well. Davis agrees. The people he means are projects that are not well funded, that take five years to build what should take two, or that hold a piece of land, a large family parcel in Texas is the example, without a real way to find customers or financing.

Memory prices have gone to about five times where they were two years ago, and they have doubled in the last year. That cost was not in the old math. Chips are then up about 15 percent each, and a newer chip is already the next q

Alex Heath describes an early investment in Groq, the inference-chip company, and a $20 billion licensing deal with Nvidia. Jonathan and much of the core team went to Nvidia. The assets stayed, and Alex Davis is running them as an inference cloud. Davis says he is usually not that early. The team was willing to change its story, and Jonathan was an easy person to back.

The deal he describes is a non-exclusive license that closed very fast at the end of last year. Most investors would have distributed the proceeds. People instead went to Nvidia to help build chips, and the sites still had to serve inference for running models and for reinforcement learning. Groq had about 53 megawatts alread

Alex Davis says Disruptive had been in Palantir for about 10 years when COVID arrived. While other people were sheltering in place, Palantir was planning to go public, which he took as a very Palantir way to think. A chance to bring in a couple of large investors, including a big insurance company, was the moment they doubled down. Holding a company for 10 years had not been the goal. At that moment, if he could write a check to anything, it was Palantir. He puts the company then at maybe $10 to $15 billion, and he does not want to be wrong on the figure. Looking back, he calls it a half-trillion-dollar company. They did not yet know the listing would be a direct listing. There was hardly a

Alex Davis keeps the list of companies Disruptive wants to own. Sometimes it is 20 names, sometimes 15, and he holds the list. Names come on and off. The job is to forecast which of them are the investments that matter, then follow them, including a company that is still too early and one that is about to break through. Missing a round at one price is not a reason to refuse a later round. The firm is the opposite of an investor who feels the chance was missed.

A higher valuation can be the moment for a larger check. The business has become something they can underwrite, or there is more they can do to take the company to market. Pitch decks are not how the good investments arrived. The weak

Alex Heath asks why leave deal-by-deal vehicles, where limited partners could see each company, for a fund. Alex Davis says this is the first real fund. Large special-purpose vehicles were a different market in 2012. A decade of groups that call themselves family offices, some with hundreds of investors, has made it hard for a company to know who is actually on the cap table.

A company that is 10 times oversubscribed, and that could take $10 billion, would rather take the money from a discretionary fund than from vehicles it cannot identify. He also treats the structure as a regulatory problem. Managing over $150 million is one line at which a firm can be an exempt reporting advisor, and so

Alex Davis says the firm has already been writing some of the largest checks in growth-stage companies, especially over the last five or six years, after 15 years at it. A fund is the more fixed version of that habit. The rule is a set number of checks each year, and a check means a company. Cut the list to 10 of the best companies they think they can help, divide the fund by that count, and the average check is large enough that $10 billion is a coherent pool.

He treats $10 billion as the goal and describes a range of $5 billion to $10 billion. Limited partners leaned toward the top of the range, as long as the firm keeps finding outlier companies and keeps doing the work behind the scenes

Alex Davis says Disruptive was not built as a fee business by people who had left banking for venture. The question was whether private companies that already looked headed for the public market could be bought before the listing, first for friends and a few institutions, and whether that could be a firm. What stayed true, in 2012 and again over the last year, is a short list of companies that are alone in their category, and the demand to own them early.

Amazon, about 25 years ago, is his picture of the point. He remembers it then as a $300 million private company with a $60 million public float. It looked expensive. It is a $3 trillion business now. A basis that starts before a company li

Alex Heath asks why announce a $10 billion growth fund when valuations look frothy. Alex Davis calls this a market of haves and have-nots, and the right time to back a few companies rather than many. Around 2011 he remembers a total of about $10 billion of venture-backed private companies, before anyone called them unicorns. He puts the count now at 5,000 to 6,000, and says it has gone parabolic.

The firm does not try to cover every pitch. The opening is the top half-dozen to 20 companies, the ones that stayed private and still need capital. A company that stands alone in its category can deserve $5 billion, $10 billion, or maybe $100 billion while it is still private. That is not a plan to

Alex Davis does not point to regrets in what Disruptive did. Fifteen years later he does know answers he did not have then. This stretch, with the firm in the middle of what is being built, is not a stretch they have been in before. He would have doubled down on the closed models and supported them with more compute. He says the same about defense.

The rest of the record is one he is content with. The team is in good shape, investors are happy with what the firm has done, and he sleeps well. The next couple of years are for working out what the next decade looks like. He calls it the most fun job in the world, one that does not feel like going to work. The hours are long because the compani

Tobi Lütke says a shopping assistant that gets paid to choose products is a breach of trust. People want agents that work for them. If the assistant pockets something like air miles for steering the order, that trust is broken. Shopify is in active conversations about more socially acceptable ways to make the connection, because merchants still have to reach the people who will buy.

He does not want AI to pile sales onto the largest companies. Department-store merchandising worked that way: the biggest packaged-goods company got the shelf, and then it got the sales. The internet, and Shopify, broke that pattern, so a product did not have to work for everyone. It could gather a thousand true

Tobi Lütke says Shopify wants entrepreneurship to be more common and easier to reach, closer to a plan A than only a plan B. The number one moment for starting a business on Shopify is the lunch break in the local time zone. People sign up then to try a career plan they have been thinking about, after a bad meeting, news that someone is losing a job, a friend leaving, or some other dissatisfaction.

They may learn something. They may catch lightning in a bottle if other people want to place orders. Reaching for independence can turn into a business, and it does not have to be a large one. It can be exactly the thing they wanted to have. That is the outcome he wants to see more of.

Tobi Lütke is fatalistic about who pays for AI shopping agents. Alex Heath put Shopify's take rate at about 3 percent and asked whether any of it would go to the agents. Tobi Lütke says most of that already goes to the credit card companies, so the margin has to be shared and little is left to fight over or to give away.

He thinks the right move for those startups, if they can get the financing, may be to charge the person using the agent. The agents are of tremendous value. On a reverse test, the money someone would have to be paid to agree they can never use Muse again is going up by the day. A charge depends on how much unique value the product provides, whether it showed up in the right

Tobi Lütke says that once he is convinced something has to happen, he can do what a professional manager hired into the job cannot. He sent a company-wide memo requiring reflexive AI use for everything at Shopify, and he would ask, every time someone needed something from him, how their AI experiment had failed. No committee reviewed the memo before it went out. He is down on committees, and he says it might not have gone out if one had.

The memo drew headlines. Inside Shopify it did not feel controversial. Tobi Lütke says more people paid attention after the press made a big deal of it, including his mom, and that gave him a chance to talk about the AI work already underway. Alex Heath sai

Tobi Lütke says people are buying more through AI agents, and the change that matters is the kind of purchase, not the growth rate. Shopify entered the year from a fairly low base, so the percentages look absurd and do not mean much. What is increasing is project-based shopping: a deeply considered order with a fairly high value, such as everything needed to set up a home gym.

The planning can start from a picture of a room or a floor plan. An agent then works out what should be in the space. He describes a weekend project with his kids and a 3D printer, using magnets, filament, and a little electronics. The agent figures out the parts and places the order from local stores so everything is

Tobi Lütke runs a completely local Claw in his house, on a rack like the one Lulu and Christian built, with one of Dell's GB300s inside, and the rack is fully training models. As a public company CEO, sending agent work out to a lab is harder because lots of things cannot be sent there, so the work runs local. The rack also holds the historical corpus of Shopify.

He is an atoms minimalist and a digital archiving maximalist. He has everything he wrote and lots of logs, and he knows what he has done on his computer for the last 15 years. Given the last 20 years, a lot about him is already in the pre-trains, so agents already know about him and his experience using them is close to ideal.

Whe

Tobi Lütke says a successful company has to be too good to ignore, and a fantastical career comes from skills and aptitudes that make a person too good to ignore. The standard is the best advice anyone can have. The company also has to be so important to people that it is deserving of a value.

The same bar applies when a product competes for the limited resources people already spend. The example, offered with some uncertainty about the exact trade, is someone stopping the extra cup at Starbucks once a month so they can afford the product. In the West, a consumer company is really competing with spending on coffee. Enterprise buying is different. Alex Heath, pricing a newsletter a year ago

Tobi Lütke says Europe is more cynical about AI than America, not simply more negative, because people there still reason from an early ChatGPT install while America keeps adopting early.

Business formation is up, and attitudes toward AI lead in the United States. Other parts of the Western world sit on an early incline of America's slope, and the old societal stereotypes show up in the data. Time spent in Europe recently sharpened the contrast: the mood is more cynical, and definitely German. Waiting until technology proves itself is fair for a country to choose. In conversation, people say they understand AI because they installed ChatGPT when it came out and it hallucinates a lot. Five m

Tobi Lütke says more Shopify storefronts are being opened now than ever, and the curve is going really, really well. The people opening them are younger. The new generation is extremely entrepreneurial.

The pattern combines AI use with the media young people already watch. Parents in every generation malign the media consumption of the next generation, and the complaints come with no break. He does not know how often Mr. Beast got banned in various households, and he is not sure that was the right call, because the channel is clearly entertainment and very, very entrepreneurial by itself.