After several days of silence, OpenAI CEO Sam Altman has finally collected himself and spoken out about Chinese AI startup DeepSeek throwing Silicon Valley into sheer chaos.

The company's R1 AI model rattled investors this week, demonstrating that it can run circles around competing models from the likes of OpenAI, Meta, and Anthropic — at a tiny fraction of the cost of training infrastructure.

After the resulting bloodbath — a more than $1 trillion tech wipeout — Altman presented moderately conflicting messages: that he was impressed by DeepSeep, and that he intends for OpenAI to deliver an epic beatdown on it.

"Deepseek's R1 is an impressive model, particularly around what they're able to deliver for the price," Altman tweeted Monday evening. "We will obviously deliver much better models and also it's legit invigorating to have a new competitor!"

Instead of tapping into the promises of lowering computing costs, though, Altman is unwilling to take his foot off the gas.

"But mostly we are excited to continue to execute on our research roadmap and believe more compute is more important now than ever before to succeed at our mission," he added in an apparent attempt to reassure spooked investors balking at rocketing capital expenditures.

"The world is going to want to use a LOT of AI, and really be quite amazed by the next-gen models coming," he added.

But whether that's the kind of rallying cry investors wanted to hear remains to be seen. DeepSeek impressed with its ability to deliver a similar AI chatbot experience while relying on far less compute. Altman is arguing that even more compute could bring about better AI, while DeepSeek's value proposition could prove a much easier pill to swallow in the short run. After all, the sky-high costs — and environmental toll — associated with generative AI have eclipsed any promises of a return on investment with a quick turnaround.

"The winners won’t be the ones burning the most cash," AI enterprise platform Cohere founder Aidan Gomez told the Financial Times. Instead, he said, it'll be the ones who find the most "efficient solutions."

How OpenAI will respond to DeepSeek's challenge and what exactly Altman meant by "much better models" remains to be seen.

Just last week, Altman encouraged his three million X-formerly-Twitter followers to get a grip, likely responding to swirling rumors that the AI industry was set to "announce a next-level breakthrough that unleashes PhD-level super-agents to do complex human tasks."

"Twitter hype is out of control again," he wrote in a January 20 tweet, just days before DeepSeek took the tech world by storm. "We are not gonna deploy [artificial general intelligence] next month, nor have we built it," referring to an ill-defined term that describes the point at which an AI can surpass human cognitive capabilities.

"We have some very cool stuff for you but pls chill and cut your expectations 100x!" he added.

Given the sheer amount of capital being used to train OpenAI's increasingly insatiable AI models, it's a high-stakes game. Altman appeared alongside president Donald Trump last week, during his announcement of a $500 billion AI infrastructure initiative dubbed Stargate. OpenAI is expected to commit $19 billion to the venture.

That kind of major dedication to the resource-heavy paradigm could have investors asking some difficult questions now that DeepSeek has demonstrated that similar AI models can be trained for a fraction of the cost.

In his Monday statement, Altman stuck to his guns, reiterating that OpenAI's number one goal is to realize AGI.

"Look forward to bringing you all AGI and beyond," he added in a follow-up, a notable change in tone compared to his comments from last week.

More on DeepSeek: OpenAI Developer Seethes at Success of DeepSeek


Share This Article