I've been thinking about which tech companies will survive the next decade. The answer is uncomfortable: almost none of them.

Not because they'll fail in the traditional sense. Because they'll become irrelevant.

The Moat Problem

The concept of a "moat" in business is borrowed from medieval castles. A moat protects you from competitors. It's what makes your business defensible.

Traditional tech moats included:

  • Network effects: The more users, the more valuable (Facebook, LinkedIn)
  • Switching costs: Too painful to leave (Salesforce, enterprise software)
  • Data advantages: Proprietary data creates better products (Google)
  • Technical complexity: Hard to replicate (historically, search engines)
  • Brand and trust: People choose you over alternatives (Apple)

Here's the problem: AI is filling in most of these moats.

Network effects? AI can simulate network value. Switching costs? AI makes migration trivial. Data advantages? Foundation models trained on internet-scale data make proprietary datasets less valuable. Technical complexity? AI democratizes engineering capability.

The moats that protected tech companies for two decades are evaporating.

The Execution Moat Illusion

"But we execute better than anyone," say a thousand startup founders.

Execution used to be a moat. Building software was hard. Recruiting engineers was hard. Shipping reliably was hard. Being good at hard things created defensibility.

AI is making execution easy.

What once required a team of engineers, designers, and product managers can increasingly be done by a single person with AI tools. The 10x engineer is becoming the 100x engineer. Soon it'll be the 1000x engineer.

When execution becomes trivially easy, execution is no longer a moat. It's table stakes.

The Feature Parity Problem

Consider any SaaS company. They've built features over years. Accumulated technical debt. Developed institutional knowledge.

An AI-native competitor can replicate their feature set in weeks. Not perfectly—but well enough. And then iterate faster because they're not carrying legacy baggage.

This isn't hypothetical. It's happening now. Every SaaS feature that can be described in natural language can be built by AI in natural language.

The features you spent five years developing? Six months of runway for a competitor with AI tools.

The Aggregation Trap

Many tech companies are aggregators. They sit between suppliers and consumers, capturing margin from the transaction.

But aggregation depends on information asymmetry. Consumers use aggregators because finding and comparing options is hard.

AI removes information asymmetry. An AI agent can search, compare, negotiate, and purchase across the entire market in seconds. Why use an aggregator when your AI can do better aggregation?

Travel sites, price comparison engines, review aggregators—all face existential challenges. Not from other aggregators, but from AI that makes aggregation a commodity.

The Workflow Automation Reality

"But we're a workflow tool. We automate complex processes."

Congratulations, you've built something AI does natively.

Workflow automation companies essentially encode human decision-making into software. If-this-then-that logic. Approval flows. Data routing.

AI doesn't need encoded workflows. It understands intent and figures out the workflow dynamically. Your carefully constructed automation is a rigid approximation of what AI does flexibly.

The entire category of workflow automation software is headed for obsolescence. Not immediately—but faster than most investors expect.

Community Is the Only Moat

So what survives?

I keep coming back to one answer: community.

Not community as a marketing buzzword. Community as a genuine human connection that AI cannot replicate, simulate, or disintermediate.

Reddit might survive because the value isn't in the software—it's in the humans posting. Discord might survive because the value is in the relationships, not the features. Substack might survive because readers subscribe to writers, not platforms.

But notice the pattern: these are platforms for human connection, not products that deliver value independently.

If your product's value can be delivered by AI, your product is in trouble. If your product's value comes from humans connecting with other humans, you might be okay.

The Identity Layer

There's one other potential moat: identity and trust.

When AI can generate anything, authenticity becomes valuable. Knowing that a specific human created something has value independent of the creation itself.

This is why I think creator-focused businesses might survive. Not because creators are better than AI at making content—increasingly they're not. But because people want to connect with humans, not outputs.

The future might be less about what you make and more about that it's you making it.

The Math Is Brutal

Let me lay out what I think happens:

Survives:

  • Platforms for genuine human connection
  • Businesses with extreme network effects in human relationships
  • Infrastructure that AI runs on (compute, chips, energy)
  • Creator businesses with authentic audience relationships
  • Businesses solving physical-world problems AI can't touch

Dies:

  • SaaS that automates knowledge work
  • Aggregators of digital goods and services
  • Software tools that encode human decision-making
  • Anything that's primarily about moving or transforming information
  • Anything AI can do 80% as well at 1% of the cost

That second list covers most tech companies. Most. The 95% number isn't hyperbole—if anything, it might be conservative.

The Uncomfortable Question

If you work at a tech company, ask yourself: What does our company do that AI can't do?

Not "what does AI not do well today?" but "what can AI never do?"

If the answer is "nothing"—if every value your company provides could theoretically be provided by AI—you're on borrowed time. Maybe a lot of time. Maybe a little. But borrowed.

The companies that will matter in 2035 are the ones solving this problem now. Building moats that AI deepens rather than fills. Positioning themselves as complements to AI rather than competition.

Everyone else is running out the clock.

What This Means for You

If you're a founder: Think hard about whether your business creates value AI can't create. If it doesn't, pivot or prepare for a difficult decade.

If you're an employee: Look at your company through this lens. Are you building skills and connections that transfer to the AI-native future? Or are you going down with the ship?

If you're an investor: Most of your portfolio is probably going to zero. The question is which companies are in the small surviving percentage.

If you're a consumer: Enjoy the next few years. You're going to get a lot of free or cheap services as companies compete for attention. Then enjoy the decade after, when AI makes everything better and cheaper.

The Silver Lining

This sounds bleak. It's not meant to be.

The death of most tech companies isn't a tragedy—it's creative destruction. The value those companies created will still exist. It'll just be delivered differently. More efficiently. To more people.

The companies that die will be replaced by AI-native alternatives that do the same thing better. The humans who worked at those companies will move to roles that leverage AI rather than compete with it.

What's ending isn't value creation. What's ending is the ability to capture value through software differentiation alone. That's uncomfortable for people who've built careers on software differentiation. But it's probably good for humanity overall.

The Real Prediction

Here's what I actually believe:

By 2035, the tech industry will look nothing like today. Not "evolved from today"—fundamentally different. The dominant companies will be ones we haven't heard of, or ones that have transformed so completely they're unrecognizable.

The transformation won't be gradual. It'll happen in waves, each wave faster than the last. Some people will see it coming and position themselves accordingly. Most won't.

The 95% figure is my honest estimate of how many current tech companies will be irrelevant—not bankrupt, but irrelevant—within a decade. The ones that survive will do so because they understood that the game was changing and changed with it.

Everyone else will be a cautionary tale about moats that weren't.

Which side of this divide is your company on? Which side are you on? The time to answer honestly is now, not when the water starts rising.