The Five Companies You Can Build in 2026

An essay about entrepreneurial options

Kyle Harrison wrote a very good essay about what venture capital is and isn’t. You should read it. If you’re not an investor (and many readers of this newsletter are not), you might be surprised at the level of introspection which is happening across the broadly defined VC space right now. A flippant description of VC today might be ‘AI and defence investing’ and that’s self-evidently a problem. Arguably though, investors are only half the issue. ‘Startups’ have also become a confusing definition. I started writing a response to Kyle’s piece and it turned into something a bit longer. Mileage may vary but I think it’s a reasonably accurate view of startup options in 2026.

There are (at least) three dynamics which new startups have to contend with. First and most obviously, AI has compressed the cost and team size of building software by an order of magnitude. But beyond this, it’s dropped the barrier to building almost entirely (I am long AI for a number of reasons but especially because it will increase entrepreneurial GDP). Second, the capital markets have bifurcated: an abundance of money at the very top (consensus bets at preposterous valuations) and at the conceptual (pre-seed, early traction) stages but a lot less in the middle (basically anything else). Third, a generation of operators who went through the 2021–2023 cycle have emerged with a bunch of interesting scar tissue and a much more nuanced view of what winning at startups actually means.

I don’t think you should ever build for a capital model but one of the highest leverage decisions you’ll make as a founder is being mindful of what kind of capital you don’t want and just avoiding those meetings and ecosystems. For anyone thinking about starting a new thing in 2026, here’s five models I think are viable.

1. Bootstrapped

Definitionally the most underrated category. Founders who never raise institutional capital, who grow on revenue, who hire slowly, and who own most of what they build. What’s changed in 2026 is that the tools available to a bootstrapped team are genuinely extraordinary. We are going to see so many more of these.

The fundamental logic of bootstrapping hasn’t changed: you need a customer who will pay you more than it costs to serve them, before you run out of money. What you cannot do is confuse “bootstrapped” with “figuring it out.” Bootstrapping requires more discipline than venture-backed building, not less. You don’t have a runway; you have margin. Protect it.

I have a suspicion that bootstrapping will remain more dominant in Europe than the US. But either way, on a ten-year horizon it’s hard to see how the number of deliberately bootstrapped (or seedstrapped, more on that below) and successful companies hasn’t exploded.

2. Seedstrapped

Flippantly I tend to think that seedstrapped is just the US version of bootstrapped, similar principles but with the caveat of ‘we’ll only ever take one investment’. Anecdotally I’ve seen a lot of seedstrapping by founders who have had less than stellar experiences with VC investments in a prior company.

I don’t think this is at all a terrible approach to building although I do worry about the dynamic with any investors. Look, maybe you took some money because you genuinely planned to build a sensible company without jumping on the VC-backed hampster wheel. Or maybe this was an experimentation phase to optimise for customer discovery. Either way, I think a lot of this tends to be experimentation to find out whether the thing is real. And if it’s real, you now have a choice: keep it close and build it profitably, or use the traction to raise on better terms than you would have gotten at zero. The reality for many founders who start with this approach is that things go well and the allure of a second cheque at a much higher valuation on highly favourable terms is just too good to turn down. But that’s the point where you absolutely need to have decided on what kind of company you’re actually building.

3. VC-backed (for speed)

Venture capital is still a very useful time-compression tool for the right category of companies: genuine power law markets, genuine defensibility, genuine founder-market fit.

I’ve watched the venture-backed playbook get applied to companies where the market is fundamentally too small, or the defensibility too thin, or the timing too wrong. And I’ve watched those founders spend years of their lives optimising for the wrong metrics because the model demanded it, then realising that giving away veto rights means ‘founder-run’ doesn’t always mean ‘founder-controlled’.

In 2026, the honest question to ask before raising venture capital is: “If I don’t raise this round, what’s the worst case?” If the answer is “I have to grow slower,” you may not need venture capital. If the answer is “a better-funded competitor takes the market before I can get there,” you probably do. Venture-backed building works when the race is real. When it’s not, you’re just paying for the privilege of answering to a cap table.

4. VC-backed (for capital scale)

The capital-intensive model has had a genuine renaissance in the last three years, driven partly by geopolitical urgency and partly by a growing recognition that some of the most important problems left to solve are ones that require atoms, not bits. [Sam Cash] In these sectors (Defence, energy, biotech, hardware, advanced manufacturing etc.), capital isn’t a strategic choice, it’s a fundamental requirement. You are unlikely to vibe-code a drone company, reversible computing chip business or modular nuclear reactor in a weekend.

What makes the capital-intensive game different is that your primary currency isn’t growth metrics; it’s technical de-risking. Every dollar you spend should be buying you a cleaner answer to the question: “Is this physically possible at the scale we need, and can we do it at a cost that makes commercial sense?” If you’re spending capital and the answer to that question isn’t getting cleaner, something is wrong.

The failure mode is timeline optimism. Capital-intensive companies die when founders underestimate how long the technical milestones take and over-promise to investors (or themselves) who don’t have the patience to wait. I know a few founders in this space and they have all fundamentally committed their lives to it. Not everyone can, or is willing, to do that.

5. Roll-Up

Maybe it’s the echo chambers I sit in but roll-ups seem to be getting more popular as a startup vector. I would distinguish between a new company you start deliberately to roll up a fragmented space (or similar strategy) versus an existing company with a consolidation strategy.

Roll-up strategies sound less risky than just building a zero-to-one company. The theory is straightforward: pick a fragmented space where the incumbent businesses are subscale and ideally underinvested. You buy them at modest multiples. You install better operations, better technology, shared back-office. The group now gets valued at a larger multiple because of scale and improved operations (also more potential buyers).

AI is a genuine operational lever for roll-ups which is one of the reasons we’re seeing so much conversation about (and capital available for) them. However it requires a different skill set from any of the other models on this list. You (plural) need to be a good operator before you are a good acquirer.

Choices

But founders and funders alike would do well to step back and ask themselves, “what game are we playing?“ Because the unfortunate reality is that most of them actually genuinely don’t know (Kyle Harrison)

I’m unclear whether it’s going to be the best time in history to build a company but I feel that 2026 is certainly the easiest time I’ve ever known to start one. About a million startup commentators have made the point that you should assume you will be building a startup for a long time. Despite that, it’s still worth saying it again. And it’s not just time, irrespective of model, startups require literal life force to will into being. At the very least be deliberate about what you don’t want to do.

I gave my whole life to the streets/Then the streets went and gave about ten to the Queen (Potter Payper)