Gross Margin Isn't What It Used to Be
For traditional Software-as-a-Service (SaaS) companies, gross margins in the 70-80% range were the gold standard. That math changes dramatically with AI. The biggest difference is the 'cost of goods sold' (COGS), which for AI startups includes a significant
and recurring variable cost: inference. Every time a user queries your model, it costs you money. This is a fundamental shift from traditional software where the cost to serve one more user is near zero. VCs are no longer expecting sky-high SaaS margins from day one. Instead, they’re looking for a defensible margin, often in the 50-60% range, and a clear, credible story about how you'll improve it over time. This could involve routing simpler queries to cheaper models, optimizing your architecture, or leveraging proprietary data to reduce reliance on expensive third-party APIs. The key is to demonstrate you understand these costs and have a plan to manage them as you scale.
The LTV:CAC Ratio, Adjusted for AI
The Lifetime Value (LTV) to Customer Acquisition Cost (CAC) ratio is a classic VC metric, and it's still vital. The rule of thumb is that LTV should be at least three times the CAC. However, AI’s unique cost structure complicates the LTV calculation. Because AI gross margins are structurally lower than in traditional SaaS, the LTV is naturally compressed. A founder can't just plug their numbers into a standard SaaS formula and expect an accurate picture. VCs will dig into how you calculate LTV, specifically ensuring you're using a gross-margin-adjusted figure that accounts for those pesky inference costs. A 5:1 LTV:CAC ratio might look fantastic on paper, but if it's based on an 80% gross margin assumption when your reality is 55%, your real ratio is closer to 3:1—moving you from 'excellent' to merely 'sustainable'. Being upfront and realistic about this calculation builds credibility.
Payback Period and Cash Efficiency
With high compute costs and potentially expensive talent, cash burn is a major concern for AI startups. This is why VCs are laser-focused on the CAC Payback Period—the number of months it takes to earn back the money you spent to acquire a new customer. In a world of high burn rates, a shorter payback period means a more efficient, resilient business. Investors want to see a clear path to getting their money back from each customer, ideally within 12 months. This metric is a direct indicator of your go-to-market efficiency and pricing power. It shows you're not just buying growth with venture dollars but are building a fundamentally sound business engine that can eventually sustain itself. A company that can demonstrate a shrinking payback period as it scales is showing signs of strong product-market fit and operational discipline.
Usage, Retention, and Stickiness
In the early stages, VCs may be more lenient on revenue figures if they see rabid user engagement and strong retention. They aren't just looking at daily active users; they want to see usage depth. Are customers using your product for mission-critical workflows, or are they just experimenting? High retention cohorts prove that your product is becoming 'sticky' and indispensable. For AI companies, this is especially important. Because the underlying models can often be commoditized, your defensibility comes from being deeply embedded in a customer's workflow. VCs will look at metrics like Net Revenue Retention (NRR), which shows how much your revenue from existing customers is growing over time. An NRR over 100% means your existing customers are spending more with you, a powerful sign of a healthy, scaling business.
A Defensible Data Moat
What stops a competitor—or even the foundation model provider you’re built on—from launching a similar feature and wiping out your business? This question is top of mind for every investor. The best answer lies in building a defensible 'moat', and in AI, the most durable moat is often proprietary data. VCs want to see how your product creates a data flywheel: the more users interact with your product, the more unique data you gather. This data can then be used to fine-tune your models, improve the product, and create a better user experience that attracts more users, starting the cycle anew. This is far more compelling than simply being a 'thin wrapper' around an existing API. Investors are funding businesses, not just features, and they need to see a long-term competitive advantage that can't be easily replicated.
