I see it all the time. A founder, buzzing with excitement, slides into my DMs or corners me at a conference. They’re convinced they’ve built the next world-changing AI. They have a slick algorithm, a polished demo, and a jaw-dropping market size number they’ve ripped straight from a Gartner or Forrester report.
“The global AI market is projected to be $1.5 trillion by 2030,” they’ll say, eyes wide with ambition. “We just need to capture 0.01% of that to be a unicorn!”
And I have to be the one to pour cold water on their dreams. I have to tell them they’re dead wrong.
That top-down market sizing? It’s a fantasy. It’s a lazy, intellectually dishonest way to justify a big valuation without doing the real, grinding work of building a business. As an angel investor who has written over 180 checks to startups, including AI pioneers like Anthropic, OpenAI, Scale AI, and Hugging Face, I can tell you that a top-down TAM slide is one of the fastest ways to get a “no” from me. It signals a fundamental misunderstanding of how markets are actually won, especially in the brutal, fast-paced arena of artificial intelligence.
The Top-Down Trap: A Fantasy of Scale
Think about it. Claiming you’ll capture a tiny fraction of a multi-trillion dollar market is like saying your plan to get rich is to find a single sunken treasure chest in the entire Pacific Ocean. It’s a meaningless statement. It tells me nothing about who your customer is, how you’ll reach them, why they’ll buy from you, or how you’ll defend against the competition that will inevitably come for you.
It’s a vanity metric. It impresses no one who has actually been in the trenches, building companies from the ground up. When I was building my last company, RemoteTeam, we didn’t start by looking at the global HR software market, a figure that probably runs into the hundreds of billions. That would have been useless.
Instead, we started by talking to a dozen companies, then two dozen, then fifty. We found companies struggling with the messy reality of managing international teams—payroll, compliance, and benefits across different countries. We built a solution for their specific, urgent problems. We obsessed over their workflows, their pain points, their frustrations. That’s the bottom-up approach. It starts with a real, tangible, painful problem and builds relentlessly outwards from there. It’s how we grew and got acquired by Gusto.
The Power of Bottom-Up: From Problem to Market
The bottom-up approach is about one thing: specificity. It forces you to stop dreaming about abstract market share and start focusing on the only thing that matters: the customer.
Here’s the simple, no-nonsense framework I tell my founders to use:
Nail Your Ideal Customer Profile (ICP): Who, exactly, are you selling to? Don't say “small businesses.” That’s not an ICP. Say, “US-based dental practices with 3-10 employees that are still using paper-based appointment scheduling.” Now that’s an ICP. You can visualize them, you can find them, and you can talk to them.
Count Them: How many of these dental practices exist in the US? A quick search on industry association websites or government data portals can give you a pretty accurate number. Let’s say it’s 50,000. Now you have a real, countable number of potential customers.
Price Your Solution Based on Value: Don’t just look at your costs and add a margin. Figure out how much value you’re creating. If your AI-powered scheduling tool saves a dental practice 10 hours of administrative work per week, what’s that worth? If it reduces no-shows by 20%, what’s the revenue impact? Your price should be a fraction of that value. Let’s say you charge $100 per month, or $1,200 per year.
Do the Math: Now you can calculate your Total Addressable Market (TAM). It’s 50,000 dental practices multiplied by $1,200 per year. That’s a $60 million TAM. Is it a trillion-dollar market? No. But it’s a real, believable market that you can actually go after and dominate. That’s infinitely more attractive to an investor than a fantasy number.
The 5 True Competitive Moats for AI Startups
This brings me to my next point. Once you’ve identified a real market, how do you win it and defend it? I’ll give you a hint: it’s not your algorithm.
In the age of open-source models and powerful APIs from companies I’ve invested in like OpenAI and Anthropic, your algorithm is a commodity. It’s a starting point, not a lasting defensible advantage. Someone else will always have a slightly better model, a slightly faster inference time. Relying on your algorithm as a moat is like building a castle on quicksand.
So what are the real moats? After looking at thousands of companies, I’ve found there are five that actually provide long-term defensibility:
1. Proprietary Data
This is the holy grail. If you have a unique, valuable dataset that no one else can access, you have a powerful, compounding advantage. Your AI models get better as you get more data, which attracts more users, who generate more data. It’s a virtuous cycle. At my first company, MovieLaLa, we collected millions of user ratings and reviews. This proprietary data allowed us to build a recommendation engine that was far more accurate than anything our competitors could create. They could copy our UI, but they couldn’t copy our data.
2. Distribution
How do you get your product into the hands of your customers? A unique, scalable, and defensible distribution channel is a massive moat. This could be a viral loop built into the product, a strong brand that drives organic traffic, or a strategic partnership that gives you exclusive access to a large customer base. RemoteTeam, for example, built a strong distribution channel through content marketing. We wrote extensively about the challenges of remote work, which attracted a large audience of potential customers.
3. Brand
A strong brand is an incredibly powerful moat, yet it’s often overlooked by technical founders. Brand is not just about a logo or a catchy name. It’s the emotional connection you build with your customers. It’s the reason people line up for a new iPhone, even when there are cheaper and more powerful phones on the market. It’s about trust, reputation, and community. Building a brand takes time and consistency, but it’s one of the most durable moats you can create.
4. Network Effects
This is when your product or service becomes more valuable as more people use it. The classic example is a telephone network. A single phone is useless, but a network of a billion phones is incredibly valuable. In AI, network effects can be more subtle. For example, a platform that connects AI experts with companies that need their services has a two-sided network effect. The more experts on the platform, the more valuable it is for companies, and vice versa. These are incredibly difficult moats for competitors to overcome.
5. Ecosystem & High Switching Costs
This is about building a platform, not just a product. Can other companies build on top of your solution? Do you have APIs and integrations that embed your product deep into your customers’ workflows? The more you become an integral part of their operations, the harder it is for them to leave. Think about the Salesforce AppExchange or the Shopify App Store. These ecosystems create incredibly high switching costs. Even if a competitor comes along with a slightly better or cheaper product, the pain of ripping out the incumbent and retraining the team is just too high.
Stop Chasing Unicorns, Start Solving Problems
Look, I get it. The allure of a trillion-dollar market is strong. It’s tempting to believe that you’re just one brilliant algorithm away from building the next Google. But I’m telling you, that’s a lottery ticket, not a business strategy.
Building an iconic AI company isn’t about chasing fantasy numbers. It’s about being disciplined, focused, and relentlessly dedicated to solving a real problem for a specific customer. It’s about the unglamorous work of customer interviews, bottom-up market analysis, and the slow, patient process of building a real, defensible moat.
So, the next time you’re working on your pitch deck, do yourself a favor. Delete the top-down TAM slide. Instead, show me your bottom-up analysis. Show me you’ve done the hard work to understand your customer. Show me you have a plan to build one of the five true moats. Show me you’re a problem-solver, not a dreamer. That’s what will get my attention. That’s what will get you funded.
Frequently Asked Questions
What factors matter most in this comparison?
For most founders, the three factors that matter most are: total cost of ownership, ease of implementation, and how well it integrates with your existing workflow. Features are important but often overweighted in decision-making.
Can I switch later if I make the wrong choice?
In most cases, yes. The switching cost is usually lower than people fear. The bigger risk is analysis paralysis, spending months evaluating options instead of picking one and learning from real usage.
Which option is best for startups?
It depends on your stage, budget, and specific needs. Early-stage startups should prioritize flexibility and low cost. Growth-stage companies can afford to optimize for performance and scalability. There's no universal answer.