Hello friends, and welcome to The Unsophisticated Investor! Brought to you by Scott & Rob, the founders of Shuttle!

If you want to invest alongside the VC funds who've backed breakout companies like Revolut, Asana, JustEat, Bolt, Lets Get Checked, Loom, Runna, Charlotte Tilbury, Deel, Aircall, AngelList, Carta, TransferWise and many more, regardless of your knowledge, network or net worth, join our limited waitlist now.

Now, let’s get into it 👇

Finding moats in the AI age

When everything is a product, the moat is permission.

A quick note before we get into this week's edition. Starcloud, which we featured a few weeks ago as the YC-backed startup that launched the first Nvidia H100 GPU into orbit, has just become the fastest company in Y Combinator's history to reach unicorn status. The company raised a $170 million Series A at a $1.1bn valuation, just 17 months after its demo day, in a round led by Benchmark and EQT Ventures that brings total capital raised to $200 million. Nebular, the fund run by Finn Murphy, was a first check investor in Starcloud and followed-on his investment in the latest round. 

The team has shown just how fast a private company can compound. We will have more on what this means for the orbital compute thesis in a future edition. For now, it is simply worth noting: the private markets moved first, and they moved fast.

This week, we want to zoom out from space and ask a question that is quietly reshaping how the best founders and investors think about building companies in any sector: in a world where AI is compressing the cost of writing software towards zero, what actually constitutes a defensible business?

The honest answer, we think, is not what the last decade of VC investing suggested.

The moat was always the friction

For much of the 2010s, venture capital rewarded a particular kind of company: one that used technology to strip friction out of a market. Better UX, faster onboarding, mobile-first everything. The model was clean. You built a product that was meaningfully easier to use than the incumbent, you grew quickly, and the combination of network effects and switching costs created a moat over time.

AI does not break this model. It accelerates it. If anything, AI makes it cheaper and faster to build the kind of polished product surface that used to require large engineering teams. Which means that the competitive advantage that came from building something beautiful and functional is shrinking rapidly. Product quality is becoming table stakes.

This is a problem if your moat was always in the code. And for a significant portion of the fintech sector, it was.

Why fintech is the clearest case study

Fintech is worth dwelling on here because it sits at an interesting fault line. The sector built its reputation on the tension between two things: the "fin" side, which involves regulation, compliance, credit risk and money movement, and the "tech" side, which involves product, engineering and user experience.

For most of the last decade, the market rewarded companies for leaning as far toward "tech" as possible. Regulatory relationships were something you rented from a partner bank. Compliance was outsourced. Credit risk was passed upstream. The logic was that financial services generated massive gross profit but was valued conservatively, while software was valued at a premium. If you could look like a software company while doing financial services economics, you captured the arbitrage.

Then AI arrived, and the arbitrage broke. Software multiples compressed. Companies that had positioned themselves as technology businesses found their valuations re-rated toward the less glamorous end of the spectrum. And the lesson is now becoming clear, the companies that held up were the ones that had spent years doing the hard work of the "fin" side. The ones that had actually earned their regulatory permissions, accumulated proprietary transaction data, and built their own underwriting models rather than renting someone else's.

The reason is structural. When AI makes code cheaper to produce, it compresses the value of anything that can be replicated in code. What it cannot compress is the cost of a banking licence in 50 US states or across the European Union, or three years of underwriting data that tells you how a specific customer cohort behaves under stress, or the institutional relationship with an FCA authorisation. Those things are expensive to acquire, take years to build, and cannot be reproduced by a better engineer with a faster laptop.

In other words: the moat was never really in the product. It was in the friction that the "fin" side imposed. The companies that treated that friction as an obstacle to be routed around are now the most exposed. The ones that treated it as a structural advantage to be earned are, quietly, in a much stronger position.

This flips the conventional framing almost completely. Compliance overhead that looked like a cost centre was actually an asset accumulation strategy. Regulatory permissions that looked like a drag on speed to market were actually barriers to entry being constructed, slowly, in public. The smart fintechs understood this intuitively. Many of the others are only realising it now.

What this means beyond fintech

The principle is not unique to financial services. It applies to any sector where regulation, proprietary data, or institutional trust creates real barriers, and where the temptation has been to treat those barriers as a temporary inconvenience rather than a strategic foundation.

Healthcare. Legal services. Infrastructure. Energy. Any market where the difficult bit is not building the product but earning the right to operate is a market where AI changes the calculus in favour of the companies that did the hard work first.

The broader argument is this: AI rewards depth over surface. If your competitive advantage is visible in the product, it can be replicated. If it lives in something that cannot be written in an afternoon by a capable engineer with a good prompt, it holds. Proprietary data. Regulatory permission. Embedded institutional relationships. Hard-won track records. These are the things that compound in an AI world rather than compress.

Building the moat, not renting it

We are in the process of securing our FCA regulatory permissions, which means we have had reason to think carefully about exactly this question.

The easy version of what we do is a platform that connects investors to private market deals. And that is accurate as far as it goes. But the FCA authorisation process forces you to think about what you are actually building at a deeper level: who your customers are, how you assess their suitability, how you handle conflicts of interest, what your compliance infrastructure looks like, how you demonstrate that your processes are robust across a range of market conditions.

It is slow and expensive in the near term. But what it produces, on the other side, is a set of permissions and institutional credibilities that a well-funded competitor cannot simply replicate by hiring more engineers. Regulatory permission is not a feature. It is infrastructure. And infrastructure takes time to build.

The other thing the "fin" framing clarifies is the role of data. Every deal that runs through Shuttle, every investor that engages with a company on the platform, every piece of diligence that gets done, generates signal about how private market participants actually behave. That signal accumulates. And over time, the platform that has seen the most deal flow, across the most investor types, in the most varied market conditions, has a genuine informational advantage that no amount of clever code can substitute for.

We are still early. But the direction is deliberate.

The bottom line

The age of AI does not make defensible businesses rarer. It makes them more important and more distinguishable. When product quality is cheap to achieve, the gap between companies with real structural advantages and companies with good products widens rather than narrows.

The clearest version of this lesson is playing out in fintech right now. The companies that earned the friction, that accumulated the data, that held the regulatory permissions, are the ones that look best positioned as AI restructures the cost base of everything around them.

The ones that rented the "fin" side to look more like "tech" are finding that the arbitrage no longer works in the way it once did.

What we’ve been working on at Shuttle

• Getting our latest drop ready for release 🎯 

• Secured our first commitments ahead of our fundraising round 💪

• Reviewing two more deals for the pipeline 🔍

The Unsophisticated Investor is brought to you by Scott & Rob, the founders of Shuttle. We’re both sick of private markets being a playground exclusive to the ultra-wealthy so we started a company to challenge the status-quo. Shuttle’s singular focus is to unlock private markets for Millennial and Gen Z tech professionals and help them build wealth through the highest performing private market opportunities.

Scott & Rob
Shuttle Co-Founders