SYMBOL – 25 January 2025, Baden-Württemberg, Rottweil: A man operates an iPhone on which apps can be seen. Photo: Silas Stein/dpa (Photo by Silas Stein/picture alliance via Getty Images)
dpa/picture alliance via Getty Images
Clay turned $1 million in annual recurring revenue into $100 million in two years and now carries a $3.1 billion valuation. It does not train a frontier model. Neither does Lovable, which reached $100 million ARR in eight months by wiring together intelligence it rents from OpenAI and Anthropic. The model is no longer the product and it certainly isn’t the moat, when all companies are wrappers the real moat is GTM and distribution.
Venture capital has already priced this in. When Lovable raised a $330 million Series B in December 2025 at a $6.6 billion valuation, Menlo Ventures partner Matt Murphy described the thesis plainly: the company built a beloved layer on top of the labs’ models that customers want to pay for. CapitalG, Alphabet’s growth fund, led that round and also led Clay’s Series C at a $3.1 billion valuation. The smart money is not buying weights. At roughly 33x ARR for Lovable, a revenue multiple that would have looked absurd in 2021, investors are underwriting distribution, retention, and brand, because the underlying capability is available to any competitor with an API key.
The commoditization is here; Lovable, Cursor, and thousands of smaller products call the same model providers, so feature parity arrives within weeks of any launch. Sequoia’s 2025 outlook argued the AI race would turn on enterprise distribution and consumer mindshare rather than raw model quality, and the revenue data supports the claim. Cursor raised $2.3 billion in November 2025 at a $29.3 billion valuation on the strength of usage, not a proprietary model. The capital is flowing to the labs, not the layers above them. Anthropic alone took in roughly $13 billion in a single quarter of 2025, and that war chest funds the very models any competitor can license through an endpoint, which forces differentiation somewhere other than capability.
That reframes go-to-market as the scarce asset, and the new playbook runs lean. Lovable reached nine figures of revenue with 45 employees and no paid acquisition budget, relying on community virality and a product engineered to be shared. Clay built the opposite motion for outbound. Its platform automates research, enrichment, and personalized sequencing, and it created a job title to operate the machine. Clay now points to more than 280 GTM engineer roles posted across companies including Cursor, Webflow, and Notion, with independent bootcamps graduating thousands of practitioners. The motion works because the product proves its own value fast. At Zendesk, teams using Lovable moved from idea to working prototype in three hours instead of six weeks, the kind of utility that turns users into evangelists and feeds the loop without a sales call.
Clay’s enterprise net retention sits above 200 percent, which means existing accounts expand faster than the company has to replace them. Its customer list includes OpenAI and Anthropic, the same labs whose models power the wrapper economy now buying the tooling to sell their own products. The loop is recursive; the labs commoditize software creation, the applications built on top compete on distribution, and the GTM layer that wins that fight becomes the most defensible business in the stack.
Skepticism is palpable and warranted on the multiples. Lovable’s paid revenue doubled to $200 million between July and November 2025, yet Barclays research reported a 40 percent decline in site traffic from its peak over a similar window, a sign that free-tier experimentation cooled even as serious buyers stayed. Vibe-coded software also carries documented quality and security risk, which limits how far the lean-team model stretches into regulated industries. A premium multiple assumes the distribution moat holds while a dozen funded competitors run the identical playbook.
A clever prompt or a fine-tuned variant buys a few weeks of lead. A distribution channel, a community, or an automated GTM engine buys years. The companies compounding fastest treat product-led growth and sales as a sequence rather than a choice, and they staff go-to-market with engineers rather than closers alone. For investors, the diligence question has shifted from what model edge a startup holds to what happens to its numbers when the capability underneath turns free. The wrappers that survive will be the ones that built distribution before the edge they assumed they had quietly disappeared.