QuietCapital partner Michael Bloch recently published a long-form post, directly arguing that most companies’ understanding of a moat is already outdated. He warns that in an era when AI can copy any software and automate any process, the real moat is not about “something being hard to do,” but about “something being hard to obtain.”
(Boosting employee productivity by 10x with AI doesn’t equal increasing company value by 10x: where did the productivity go?)
AI is rewriting the definition of a corporate moat
For a long time, the tech industry has viewed “technical complexity” as the most important competitive barrier. Embedding your product deeply enough and making it take a rival a year to replace you was once a sufficiently safe strategy. But Bloch believes this line of thinking is facing a fundamental collapse.
AI is compressing the time required to “get things done,” but it cannot compress the time required for “things to happen.”
He emphasizes that AI can significantly speed up writing software and integrating features into a copied product; however, the passage of time in accumulating real users, securing government approvals, and building physical factories is an unavoidable reality that no technology can bypass.
The five moats: the most effective competitive barriers in the AI era
- Proprietary data that keeps accumulating
Bloch stresses that not all data has moat value. A static dataset that’s simply expensive to collect will eventually be synthesized or bypassed. What truly has value is “live data”—data generated continuously through day-to-day operations, and data whose production is itself difficult to replicate.
He uses an agriculture tech company, Orchard AI, as an example: the company installs cameras on farm equipment, tracking billions of fruit across multiple growing seasons and multiple geographic regions. This dataset updates every day, and each field operation makes the model smarter.
Competitors can’t catch up by training models with public data. The only way is to use the same kind of machines, go through the same orchards, and accumulate year after year.
- Network effects
Each new user increases the value of the product—this is the essence of network effects, and also one of the most difficult assets to replicate.
Bloch uses DoorDash as an example: more drivers make delivery faster, more restaurants give consumers more choices, and more consumers make the overall economic model healthier. You can easily Vibe Coding build this app, but you can’t replicate the density of drivers, restaurants, and users that has been cultivated across ten thousand cities over many years.
He also points out that AI’s widespread adoption may actually make the “cold-start problem” even more severe. When it becomes easy to build a feature-complete competing product, hundreds of alternatives will appear at once, all fighting for the same network. This will allow platforms that already have liquidity and a user base to grow with compounding returns, while everyone else can only scramble over what’s left.
(Can you make extra money just by folding clothes videos? DoorDash Tasks has couriers training AI on the side and sparks heated discussion)
- Government regulatory approvals
Government operates at the pace of politics, not the pace of technological progress. Defense tech company Anduril needs procurement approval and confidential contracts to sell to the Department of Defense; getting a bank license takes years; FDA drug approvals also take years.
Bloch notes that as AI capabilities keep improving, the scope of relevant regulations will only expand, not shrink—because the stronger the capability, the higher the necessity for regulation.
(Anthropic sues the Pentagon: being blacklisted and shut down could cause losses of tens of billions of dollars and deal a severe blow to fundraising ability)
- The ability to mobilize large-scale capital
Bloch believes this is the strongest moat in the current market. Ultimately, the competition is still in the real world: a $20 billion data center build, and a $10 billion nuclear power plant build. When the bottleneck shifts from software to physical reality, the ability to raise and deploy massive amounts of capital will become one of the most critical advantages in this era.
More importantly, the ability to mobilize capital is not just about money—it’s about institutional trust, past track records, and relationships between people, all of which take decades to build.
- Physical infrastructure
Factories, power plants, battery networks, and data centers—these tangible assets form the final and also the most difficult-to-cross moat. Bloch uses Base Power as an example: the company is deploying thousands of battery units in homes across Texas, while also building its own manufacturing facilities. Installing each unit is a tangible asset that generates revenue in a real backyard.
You might be able to design the entire system in a week with AI, but you can’t manufacture, install, and connect thousands of physical devices in a week. Physical laws impose a hard deadline on the timeline, and no AI can break through it. The advantage of whoever starts building first will keep widening over time.
What moat is disappearing?
Bloch also clearly identifies which advantages—once seen as barriers—are quickly losing their meaning: “Deep embedding and integration between products sounds solid, but the cost of swapping it out is, in the end, just engineers’ work hours—which is exactly what AI is best at compressing.”
He points out that ecosystem lock-in used to be like a fortress, but once AI can rebuild system integration at the speed of describing the problem, the wall is no longer sturdy. Software-scale advantages, in a world where engineering costs approach zero, have lost their significance.
These are moats built to fight “intelligence scarcity,” but intelligence is no longer scarce.
“Time” is the ultimate moat
Behind the five moats is a shared underlying logic: they all require time in the real world that cannot be compressed in order to accumulate. Network density needs years of user adoption; regulatory approvals require years of political processes; infrastructure needs years of construction; data needs years of field operations; capital relationships require decades of trust accumulation.
“Time that can’t be parallelized is the foundation under all five moats.” Bloch says that companies that already have these moats are widening the lead every day, because first-mover advantage itself is a moat.
For investors and founders, Bloch leaves a core question: “Is your competitive advantage based on others being unable to ‘build it,’ or based on others being unable to ‘build it in time’?” In the AI era, this answer can determine who can survive.
Which companies can survive in the AI wave? Venture capital partners name the five moats: first-mover advantage matters most. First appeared in Chain News ABMedia.
