Why Were We Surprised by DeepSeek?

The only real surprise here is that anyone was surprised at all.

DeepSeek’s AI announcement—boasting OpenAI-level capabilities at a tenth of the cost—has been met with shock and endless headlines about “disruption.” But let’s be clear: this was always going to happen. China’s history in tech has shown us the same pattern repeatedly—enter the race, deliver something faster and cheaper, and claim market share by sheer efficiency. Why does this shock anyone?

It’s not just that China showed up in the generative AI space; it’s that DeepSeek has made the conversation shift overnight. Last week, AI insiders were proudly touting the astronomical costs of development as a badge of exclusivity. Now, those same players are scrambling, embarrassed by how much they’ve spent as DeepSeek pushes the narrative toward accessibility and efficiency.

But before we lose the plot, let’s take a step back. The real story here isn’t the price tag, nor the geopolitical finger-pointing. It’s about what DeepSeek’s move means for AI, for product builders, and for the future of innovation itself.

Pre-order our book! Building Rocketships: Product Management for High-Growth Companies. A Blueprint for Product-Led Companies. It’s the book for founders, product managers, and product-led executive teams that will increase the success of any company.

Why We Shouldn’t Be Surprised

Let’s start with the basics.

• This is what China does. DeepSeek’s ability to replicate the work of OpenAI and others at a fraction of the cost is a classic example of leveraging efficiency, speed, and an existing playbook. Open-source tools like LLaMA were right there, and DeepSeek knew how to build on them without starting from scratch.

• This isn’t about advancing technology. DeepSeek didn’t announce a groundbreaking new method or a technical leap. It’s about doing the same thing more efficiently, using lower-precision computations and optimization strategies that have been floating around for years.

• Hype always breeds disillusionment. The broader generative AI space has been running on unsustainable levels of hype. Every technology cycle reaches a point where expectations crash into reality. DeepSeek isn’t some apocalyptic disruptor—it’s just a reminder that the race was never going to stay exclusive to a handful of big names.

Open Questions vs. Real Implications

There are plenty of open questions about DeepSeek’s announcement. What are the geopolitical ramifications? Can we trust their technology? Is their pricing sustainable? These are all valid points—if you’re a policy maker or journalist. But for product builders, they’re distractions.

Here’s what actually matters:

1. Cheaper, Faster Inputs Are Good for Builders
   DeepSeek’s claim of AI capabilities at one-tenth the cost—whether perfectly accurate or not—means cheaper, more accessible infrastructure. That’s a win for anyone building on top of these systems. The faster and cheaper foundational models get, the more room there is for innovation at the application layer.

2. Open-Source Is Better for Builders
   DeepSeek’s choice to embrace open source was a strategic acknowledgment that this type of model will be a commodity. By inviting the world to adopt and iterate on their technology, they accelerate adoption and make their model more valuable—even at the risk of losing a competitive edge. For product teams, this openness means faster experimentation and less risk.

3. Don’t Get Distracted by the Noise
   Furthermore, the relationship between cost and adoption creates a fascinating dynamic that many overlook. While some worry about DeepSeek's pricing strategy undermining existing players, they miss a crucial economic principle: as prices decrease, adoption increases, potentially expanding the total addressable market. Yes, this might challenge the dominance of some companies and reduce barriers to entry, but it ultimately drives innovation and benefits both product builders and consumers. The question isn't whether cheaper, more accessible models will disrupt the market—it's how builders will leverage this accessibility to create more value for end-users.

   For product teams, this isn’t about geopolitics. It’s about opportunity. As the base cost of generative AI drops, the ability to build and deploy transformative applications only increases.

Computing obeys the gas law. This means, it fills the available space as defined by available resources

-Pat Gelsinger, Ex-CEO of Intel, On DeepSeek’s emergence.

What This Means for Product Builders

DeepSeek’s announcement reinforces a fundamental truth about any new technology cycle: the biggest winners aren’t the providers of inputs—they’re the builders of applications. The economics of this situation are even more compelling when we look at historical precedent. In the last two decades, the ratio of enterprise value created by silicon and its value chain versus the software it powers, has consistently landed between 2:1 and 3:1. This isn't just about margins—it's about the fundamental nature of value creation in technology. The real opportunity lies not in the models or the GPUs themselves (inputs), but in creating software that transforms workflows, enhances marketing, improves sales, and revolutionizes customer support (outcomes). The true game is, and always has been, about building applications that make these technological advances accessible and useful to businesses and end-users.

ChatGPT wasn’t transformative because of its underlying model; it was transformative because it put generative AI into the hands of everyday users and businesses while satisfying our need for a social interaction model (chat) and information-seeking (information search and assistance).

DeepSeek might be an important input, but it’s not the story. Its emergence feels disruptive now, but it will look obvious in hindsight. Of course, a new player was going to make the same tech cheaper, faster, and more accessible. Of course, the cost of generative AI was going to drop. This happens with every major technological shift, from the Industrial Revolution to the rise of the Internet.

For product builders, the takeaway is simple: don’t get distracted.

The open questions—about geopolitics, open-source licensing, or market competition—will be debated endlessly. But they’re not your problem. What matters is how you use the tools available today to build something valuable. The infrastructure will keep improving, and the costs will keep falling. The opportunity is yours to take.

Photo by Andrea Piacquadio