It's not every day that a relatively unknown player from China crashes the party hosted by tech titans like OpenAI, Google, and Anthropic—but that's exactly what DeepSeek has done. Since late 2022, we've seen a few AI assistants dominate the scene—ChatGPT, Claude, Gemini—you know the names. These tools keep getting smarter, thanks to massive cash injections (think billions), fancy data centers, and cutting-edge chips.

Then along came DeepSeek. Earlier this year, it launched its R1 model, which, according to the company, cost only around US$6 million to build. That's chump change compared to what the big players are spending. Even if some experts suspect the real cost was higher, the message is clear: high-quality GenAI might no longer need a sky-high budget.

As CFRA analyst Angelo Zino puts it, trailblazers pay the highest price. But once the path is carved out, following it becomes a lot cheaper—and faster.

At the recent HumanX AI conference in Las Vegas, Thomas Wolf from Hugging Face echoed this sentiment, saying it's getting easier (and cheaper) to release GenAI models—and frankly, people are starting to care less about which model they use. "We're heading into a multi-model world," Wolf said, noting the lukewarm response to the latest ChatGPT update.

Not All Models Are Equal – Yet

OpenAI's Chief Product Officer Kevin Weil wasn't having it, though. He countered the idea that all GenAI models are interchangeable. "That's not true," he said. While OpenAI might not have the one-year head start it once did, he believes a three-to-six-month lead still matters—and the company plans to fight hard to keep it.

With a user base of 400 million and the brand recognition that comes with it, OpenAI has what some call the "Google advantage." It's the default in people's minds, which gives it a steady stream of user data to continuously improve its AI.

Still, some experts think the playing field will eventually even out. Jeff Seibert, co-founder of Digits, believes OpenAI will maintain an edge for more advanced applications, but in many other cases, it might not matter much which model you're using. His advice to startups? Stay flexible—design your tech so you can swap out AI models easily as the market shifts.

Burning Cash, But Moving Fast

One of the key reasons behind this shift is cost. Thanks to smarter chip usage and better optimization methods, creating powerful language models is cheaper than ever.

And then there's the open-source movement—some GenAI models are being made public, allowing developers worldwide to tweak, improve, and build on them. This has sped up progress and brought more competition into the game.

Meanwhile, the big players are still burning through cash at an eye-watering rate. OpenAI, for instance, reportedly spends around a billion dollars a month to keep things running. It recently landed a US$40 billion investment from SoftBank, valuing the company at an eye-popping US$300 billion—nearly double last year's valuation.

But with that kind of burn rate, some are questioning how sustainable it all is. Jai Das from Sapphire Ventures put it bluntly: "I have a hard time seeing how they get to a point where revenues are higher than the amount of cash they burn."

Even Anthropic, the company behind Claude and a champion of responsible AI, isn't immune—it raised US$3.5 billion in March, pushing its valuation to over US$61 billion.

The Bottom Line

DeepSeek's rise is a reminder that the GenAI world is changing fast. The old guard isn't going anywhere just yet, but the gap is closing—and fast. Innovation, smart design, and adaptability might just beat out big spending. And if DeepSeek's model is anything to go by, we could be heading toward a future where powerful AI is no longer exclusive to tech giants with bottomless pockets.

Stay tuned—the AI race just got a whole lot more interesting.