AI Is Getting Cheaper. What Actually Builds Enduring Companies?

The real moat isn’t the model—it’s knowing your customers better than anyone else.

Over the weekend, I shared this thought with some of the folks closest to SuccessVP, and it sparked some great conversations—my text messages were buzzing! Given the response, I thought it might be something my broader Substack community would appreciate as well.

DeepSeek Is Dragging AI Prices Down—What Actually Drives Startup Success?

Last week, DeepSeek-R1 didn’t just break the internet—it accelerated the open-source AI revolution. While much of the discussion around DeepSeek has rightly focused on national security concerns—and there is ongoing debate over just how cost-effective this new model is—the fact remains: the price of cutting-edge AI has once again dropped significantly.

Designed to rival OpenAI’s Model O1 at a fraction of the cost, DeepSeek-R1 raises an important question: If cutting-edge AI is becoming a commodity, what truly differentiates companies? Marc Andreessen even called it a “Sputnik moment,” underscoring its significance in the ongoing AI race. Without a doubt, this marks a major milestone in the broader trend of high-quality, open-source models competing with top-tier proprietary systems. We are witnessing the democratization of AI—and founders are reaping the benefits.

A year ago, founders without a proprietary LLM were dismissed as mere “AI wrappers.” Today, some of them are the ones winning—not necessarily by building models, but by building relationships and unlocking high-quality training data—which, I’d argue, is as important in many cases than the model itself.

By focusing on customers, they’ve built deep relationships, gained valuable insights, and developed products that truly meet demand—all while leveraging competition among major LLM providers. They seamlessly switch between OpenAI, Anthropic, and Gemini, with low switching costs, constantly adding more value for their customers with every new breakthrough.

It has always been existential for startups to make their customers successful, and I’d argue that in a world where we can build faster and at a lower cost than ever before, the value of knowing your customers and truly solving their deepest needs has only increased.

Customer success is a 0-to-1 problem; founders who lean into their customers’ success are the ones who will succeed. As an investor, these founders are a source of alpha in the AI-driven world we’re moving into. I think it’s one of the reasons my recent Substack posts Founder-Led Customer Success and Why Customer Success Matters Most in a Vertical AI World have resonated so strongly.

Victor Riparbelli, CEO of Synthesia, put it best in a recent 20VC episode :

‘Sustainable growth in AI comes from ensuring long-term customer success and retention, not just aggressive new sales.’

That’s the real signal—not signing a contract, but getting a renewal. Harry Stebbings is a long-time friend; we’ve been trading notes since he was a 19-year-old with a bedroom podcast, and this conversation nails what so many are missing. They even went so far as to say that a lot of money is going to go up in flames because people are overly obsessed with technology and forgetting what matters most—customers.

At the end of the day, technology alone doesn’t build category-defining companies—customer obsession does. The startups that cultivate deep, lasting relationships won’t just survive; they’ll dominate. In a world where AI is abundant, the real moat isn’t the model—it’s knowing your customers better than anyone else.

—JG

Comments (0)

Popular