Anthropic recently said something remarkable: they're building software and solutions faster than the industry can absorb. They've disrupted coding, security, legal, financial infrastructure, and even COBOL modernization. All of this in a matter of weeks. A decade's worth of innovation compressed into seven days.
In the process, they've killed startups every week.
The Body Count Is Real
In February 2026, Ira Bodnar, founder of San Francisco-based ad automation startup Ryze, woke up to find her business obsolete. Claude and Manus AI had released features that directly competed with her core product - a tool for managing Google and Meta ads. One Claude/Manus feature and their close rate dropped from 70% to 20%."
She's not alone.
When Anthropic announced Claude's legal plugins in late January, the market reacted instantly. Thomson Reuters dropped 16% in a single day. RELX (owner of LexisNexis) fell 14%. LegalZoom crashed nearly 20%. In one week, $830 billion in market value evaporated from global software stocks.
When they released COBOL modernization capabilities through Claude Code, IBM, a 115-year-old company, saw its worst trading day in 26 years, plunging 13%.
When they introduced new security features, cybersecurity stocks tanked across the board.
This isn't hype. It's happening every week.
The Fear of Launching
Founders I speak with are paralyzed. They're afraid to launch because Claude, or another AI giant, might drop a feature tomorrow with instant distribution and kill them before they get traction. The innovation cycle has compressed so violently that building software now feels like building sandcastles at high tide.
Why Services Will Win
But here's the opportunity hiding in plain sight: implementation.
Claude can build software faster than anyone. But localizing Claude for a specific use case, integrating it securely into an enterprise, customizing it for regulatory compliance, training teams to use it effectively, that still requires expertise.
And that expertise lives today in services.
Consider the valuation gap. As of Q1 2025:
- Public SaaS companies trade at ~7.4x revenue
- IT services companies trade at a 1.3-1.4x revenue, a significant discount to SaaS multiples
Why? Because software was seen as scalable, and services were seen as linear.
But when software becomes a commodity, as Claude spins up solutions overnight, the bottleneck shifts from building to deploying.
Services become the moat.
The AI Services Era
I believe we're entering an era where IT services, or let's rebrand them to AI services, will see a resurgence. We'll see a plethora of companies scaling to millions, even billions in revenue by doing one thing well: helping organizations implement, localize, and secure AI.
Claude will make implementation easier, no doubt. But "easier" doesn't mean "trivial." Enterprises don't just plug in AI and walk away. They need:
- Security audits and compliance frameworks
- Custom workflows tailored to their operations
- Integration with legacy systems (hello, COBOL)
- Change management and training
- Ongoing maintenance and governance
These aren't product problems. They're service problems.
Claude Will Eat the World. Services Will Digest It.
I'll be bold: Software (AI) will still eat the world. But Services will digest it.
Software disruption creates a vacuum. Someone has to fill it. Not with more software, but with the human expertise required to make that software work in the messy, regulated, legacy-laden reality of actual businesses.
The SaaS era taught us to build once and sell infinitely. The AI era will teach us that infinite leverage still requires local implementation.
The gold rush is coming. But this time, it's not in building pickaxes, it's in showing people how to use them.