Key Takeaways
- • Traditional SaaS metrics and margins are obsolete in AI-native businesses. Gorkem Yurtseven, co-founder of fal, stated that "no one in AI really has 80–90% gross margins" because compute costs are material and directly tied to customer delivery, compressing margins while accelerating growth velocity beyond historical SaaS benchmarks.
- • Compute has replaced code as the primary unit of economic cost. In AI companies, cost of goods sold (COGS) functions equivalently to customer acquisition cost (CAC), fundamentally inverting traditional SaaS unit economics and requiring founders to optimize for compute efficiency rather than sales-driven growth.
- • New metrics are required to measure AI business health. Bessemer Venture Partners and industry leaders recommend tracking compute efficiency, usage retention, and growth velocity instead of the Rule of 40 and other legacy SaaS performance indicators that no longer apply to AI-native companies.
- • Sales models are being abandoned in favor of organic, velocity-driven growth. Anthropic and fal have ditched traditional quota-based sales systems in favor of "shadow targets," reflecting the unpredictable, rapid growth dynamics characteristic of AI products that scale through product-market fit rather than structured GTM motions.
- • AI founders are rewriting pricing, go-to-market, and growth strategies from first principles. Leaders from Anthropic (Kelly Loftus, GTM), Cursor (Jacob Jackson, ML), and fal (Gorkem Yurtseven) outlined a new playbook where SaaS frameworks collapse under AI's speed and scale, requiring novel business model design.
Summary
The artificial intelligence era is fundamentally breaking the SaaS playbook that defined software company scaling in the 2010s. According to Bessemer Venture Partners' "Scaling an AI Supernova" report, traditional frameworks built on predictable recurring revenue, efficient sales motions, and consistent gross margins of 80–90% no longer apply to AI-native companies. The shift reflects a step-function change in technology delivery, unit economics, and go-to-market strategy, with compute costs replacing traditional software economics as the primary driver of business model design.
At SaaStr Annual, Bessemer Partner Talia Goldberg convened three leaders building at the forefront of this transformation: Kelly Loftus (GTM leader at Anthropic), Jacob Jackson (machine learning leader at Cursor and founder of Supermaven), and Gorkem Yurtseven (co-founder of fal). Their insights revealed that AI companies operate under fundamentally different constraints. Yurtseven emphasized that compressed margins are a feature, not a bug: "The cost to serve each customer is real. Everyone has less margin—but they're growing like crazy." This dynamic—where COGS is inextricably linked to customer delivery and scales with usage—inverts traditional SaaS unit economics and requires a complete reimagining of how founders measure success.
The report identified compute efficiency as the new strategic lever replacing the code-centric optimization of prior-era SaaS. In this model, compute becomes the functional equivalent of customer acquisition cost, making infrastructure optimization inseparable from go-to-market strategy. Both Anthropic and fal have moved away from traditional quota-based sales systems toward "shadow targets," acknowledging that AI product growth is driven by organic velocity and product-market fit rather than predictable sales cycles.
Bessemer recommends that early-stage AI founders abandon Rule of 40 and similar legacy metrics in favor of compute efficiency, usage retention, and growth velocity as primary health indicators. This represents not merely an incremental evolution of SaaS but a fundamental recalibration of startup physics, where speed, scale, and margin compression define success in an environment where traditional playbooks have become obsolete.
At SaaStr Annual, Bessemer Partner Talia Goldberg convened three leaders building at the forefront of this transformation: Kelly Loftus (GTM leader at Anthropic), Jacob Jackson (machine learning leader at Cursor and founder of Supermaven), and Gorkem Yurtseven (co-founder of fal). Their insights revealed that AI companies operate under fundamentally different constraints. Yurtseven emphasized that compressed margins are a feature, not a bug: "The cost to serve each customer is real. Everyone has less margin—but they're growing like crazy." This dynamic—where COGS is inextricably linked to customer delivery and scales with usage—inverts traditional SaaS unit economics and requires a complete reimagining of how founders measure success.
The report identified compute efficiency as the new strategic lever replacing the code-centric optimization of prior-era SaaS. In this model, compute becomes the functional equivalent of customer acquisition cost, making infrastructure optimization inseparable from go-to-market strategy. Both Anthropic and fal have moved away from traditional quota-based sales systems toward "shadow targets," acknowledging that AI product growth is driven by organic velocity and product-market fit rather than predictable sales cycles.
Bessemer recommends that early-stage AI founders abandon Rule of 40 and similar legacy metrics in favor of compute efficiency, usage retention, and growth velocity as primary health indicators. This represents not merely an incremental evolution of SaaS but a fundamental recalibration of startup physics, where speed, scale, and margin compression define success in an environment where traditional playbooks have become obsolete.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.