What Anthropic's S-1 Actually Tells Us
When a company files for an IPO, the S-1 document is one of the most honest things it will ever publish. Investors demand it. So when Anthropic's filing — reported by TechCrunch — revealed that more than half of its enterprise customers spending $100,000 or more annually now use both the Claude API platform and at least one of Anthropic's own first-party applications, that's not a coincidence. It's a strategy.
The pattern should be familiar to anyone who's watched Amazon's playbook: get customers in the door on one product, expand the relationship with complementary offerings, and make the cost of leaving quietly prohibitive.
For business leaders evaluating or already using AI tools, this is important signal — not reason to panic, but reason to pay attention.
Why Stickiness Matters (And Not Just for Anthropic's Valuation)
In enterprise software, "stickiness" is a polite word for switching costs. The more your workflows, data pipelines, and internal tools depend on a specific platform, the more painful it becomes to change vendors — even if a better or cheaper alternative emerges.
Anthropics's cross-sell metric tells investors that its revenue is durable. But it tells you something different: that the company is deliberately deepening dependencies with its best customers.
This is not inherently malicious. Integrated platforms can genuinely deliver more value. But the risk calculus shifts once you're embedded. Price increases, service changes, or strategic pivots by the vendor become your operational problem.
For mid-market Canadian companies — many of whom are in the early stages of AI adoption — decisions made today about which platforms to build on will shape your options for years.
The Canadian Mid-Market Context
Canadian SMBs and mid-market firms are under real pressure to adopt AI tools quickly, driven by productivity demands, talent constraints, and competitive dynamics. Many are evaluating or actively deploying tools built on top of models like Claude, GPT-4, or Gemini.
The challenge is that most of these decisions are being made tactically — solving an immediate problem — rather than architecturally. A team adopts Claude for document summarization. Another uses an Anthropic-native app for customer support drafts. Before long, the organization's AI footprint is concentrated in one vendor's ecosystem without a deliberate choice ever being made.
This is exactly the trajectory Anthropic's S-1 data describes. And it's happening at companies that likely don't have enterprise procurement teams scrutinizing vendor strategy.
What to Do Before You're Locked In
The goal isn't to avoid AI adoption — that would be a costly mistake in the other direction. The goal is to adopt with eyes open. Here's what that looks like in practice:
Audit your current AI touchpoints. Map which tools, workflows, and data sources already depend on a specific AI vendor. You may be more concentrated than you realize.
Prioritize API-first and interoperable tools. Where possible, choose solutions that sit on top of multiple model providers or that allow you to swap the underlying model. This preserves optionality without sacrificing capability.
Read the contract. AI vendor agreements often include auto-renewal clauses, data usage provisions, and limited portability protections. Have legal review them with the same scrutiny you'd apply to any enterprise software deal.
Build internal capability, not just vendor dependency. The companies that will have the most leverage with AI vendors are those that understand enough about how the technology works to evaluate alternatives. Invest in internal literacy alongside external tools.
The Broader Lesson From Anthropic's Filing
Anthropics IPO ambitions are a reminder that AI vendors are building businesses, not just building tools. Their incentives — to grow revenue, increase retention, and expand wallet share — are legitimate and predictable. Your job as a business leader is to align your adoption strategy with your own long-term interests, not theirs.
The companies that will extract the most value from AI over the next five years won't necessarily be the earliest adopters. They'll be the ones who adopted deliberately — building flexibility into their architecture while still moving fast enough to compete.
That balance is achievable. But it requires treating AI procurement as a strategic decision, not just a software purchase.