How to price an AI app company: Lovable's ARR multiple

When an AI app company raises at a high multiple, the decision is whether the number can survive a test. Lovable announced a $400 million Series C at a $13.3 billion valuation, with Menlo Ventures as lead investor and the EQT-managed Scaleup Europe Fund as co-lead. The round gives founders and investors a clean case for a two-part check: compute the ARR multiple, then verify enterprise usage and model-cost control before accepting the premium.

The multiple should come from the latest disclosed round

Pull the latest valuation and revenue run rate. Put the new valuation next to the most recent run rate. Lovable was tracking to an annual revenue run rate close to $600 million by the end of August 2026, nearly three times the level disclosed in December. The ARR multiple is the valuation divided by that run rate, so the round becomes a number you can test instead of a story about momentum.

Compare the multiple to the growth behind it. Lovable's Series B in December valued the company at $6.6 billion, eight months before the new round. The previous price and the new price should be read against the same revenue window. If a company is paying a premium, the growth should be visible in the disclosed numbers, not only in the narrative. The test is whether the revenue line is moving fast enough to justify the price.

Use the latest disclosed run rate, not a forward promise. A run rate is an annualized snapshot of current revenue, so it can change quickly. If the company is using a future target, label it as a target. The result is a multiple tied to a disclosed period, not a projection.

Enterprise usage must show real adoption

Check whether named customers are real usage. A logo is a claim until it is tied to adoption. Employees at half of the Fortune 500 had used the platform within its first year, and that share rose to nearly two-thirds by August 2026. Lovable named Nvidia, Adidas, Hearst, and Zendesk as customers. The list should show plausible usage, not just names, and the adoption share should make the enterprise claim easier to believe.

Look for breadth, not just logos. From its November 2024 launch through August 2026, users created more than 60 million projects, and apps built on the platform received over 900 million monthly visits. Those numbers matter because they show the product is being used, not just bought. Usage breadth should connect a sales win to a durable revenue base, and it gives the enterprise claim a second line of support.

Ask whether inference costs can fall as usage grows. AI app revenue can be squeezed by the cost of running models. Lovable said it will keep post-training open-source models so the product can adapt to individual goals, work context, and working style. The statement is a lever, not proof that margins will improve. The mechanism should reduce dependence on expensive third-party inference, or give you a clear reason to discount the multiple.

The price follows the checks

Set a price range, not a single number. A multiple is a starting point, not a verdict. If the revenue run rate is moving quickly and enterprise usage is broad, the price range can be higher. If usage is narrow or model costs are still uncontrolled, the range should fall. The range should have a reason for each limit, so the next round can be judged against evidence rather than enthusiasm.

Make the decision explicit. Accept the multiple when growth, enterprise usage, and model-cost control all point the same way. Walk away when any one of the three is missing. For a founder, set the price only after the checks are documented. For an investor, write the reason for the premium into the memo, not just the number. The memo should show why the price is defensible, including the check that is weakest.