Lessons from Building an AI-Native Business
Gorkem Yurtseven, CTO and co-founder of FAL, shares hard-won insights from scaling an AI-native business. As the CEO of an $8B generative media platform, he’s navigated challenges unique to AI infrastructure—from shifting cost structures to redefining customer relationships. Here’s what founders can learn from his journey.
AI Margins: The Hidden Cost of Innovation
Traditional SaaS models promised high margins through low marginal costs. AI flips this logic. Every query, every user, demands real money to serve. GPUs and inference costs are rising, not falling, as models evolve. FAL discovered that newer video models, while in demand, are dramatically pricier than older image models. Margins don’t improve by waiting—they require pricing strategies that reflect true costs from day one.
Why Margins Matter
- Model costs outpace hardware improvements.
- Customer demand for advanced models drives expenses upward.
- Founders must price for cost-to-serve, not just growth.
When High-Usage Customers Become a Liability
In traditional SaaS, high usage equals high value. In AI, it’s a double-edged sword. Large customers consume massive inference resources, often subsidizing smaller accounts. FAL’s solution? Track wallet share—the percentage of a customer’s generative media spend flowing through their platform. This metric reshapes expansion strategies, prioritizing quality over sheer volume.
Key Takeaway
“Growth isn’t just about adding logos. It’s about deepening spend with the right accounts,” Yurtseven explains. Balancing usage and margin is critical.
Building an AI-Native Business: Key Lessons
Scaling an AI infrastructure business demands unconventional tactics:
1. Ditch Annual Quotas
At 50% quarterly growth, annual targets become obsolete. FAL shifted to monthly/quarterly goals, allowing agility in a fast-moving market. For now, they focus on target earnings without rigid quotas.
2. Hire Researchers, Not Resumes
FAL’s research team grew through a grants program: anyone could submit a project idea. If it aligned with their goals (efficient AI, finetuning, inference), they got compute time. Four hires came from this process. “Watching someone do the work tells you more than any interview,” Yurtseven notes.
3. Position as a Category
Contrary to early assumptions, image and language models serve distinct markets. FAL carved out a niche as a generative media platform, targeting buyers with clear use cases. This positioning became a moat in a crowded AI landscape.
The Metrics That Matter
FAL tracks three core metrics:
- Logo diversity: Aim for 30-35 paying companies to avoid concentration risk.
- Churn: Address attrition as an ongoing priority, not an annual review.
- Wallet share: Measure how much of a customer’s AI spend flows through your platform.
Interestingly, AI-native startups often spend more aggressively than enterprises. Focusing on this segment can drive near-term growth.
Final Takeaway
Building an AI-native business isn’t about shortcuts. It’s about embracing a new playbook: pricing for cost-to-serve, rethinking customer value, and hiring for impact. As Yurtseven puts it, “This is a different business with different rules. Treat it as a starting point, not a problem to solve.”
Want more insights like this? Join SaaStr AI Annual 2026 in May to connect with founders reshaping the AI landscape.
