OpenAI’s Altman Promises ‘Much Better’ Models After DeepSeek Reveal

OpenAI chief executive Sam Altman promises ‘much better’ AI models, fast-track release schedule after DeepSeek rattles markets

3 min
OpenAI chief executive Sam Altman at the OpenAI Hackathon in March 2018. Image credit: OpenAI
Getting your Trinity Audio player ready...

OpenAI chief executive Sam Altman called DeepSeek’s R1 artificial intelligence (AI) model “impressive” and would spur his firm’s own models to be “much better”, after the start-up shook markets earlier this week.

“DeepSeek’s R1 is an impressive model, particularly around what they’re able to deliver for the price,” Altman wrote in a social media post.

“We will obviously deliver much better models and also it’s legit invigorating to have a new competitor!”

Altman added that the company would fast-track some of its releases.

OpenAI logo displayed on a smartphone. Artificial intelligence, AI, ChatGPT
Image credit: Levart Photographer/Unsplash

Capital spending

The R1 model was released last week, days before DeepSeek’s app rose to the top of Apple’s download charts.

DeepSeek said it has developed its models for a relatively low cost, but that they deliver performance on par with those from OpenAI, Anthropic and others.

This, combined with the app’s sudden and unexpected popularity, raised questions about the billions in capital spending by Western tech firms over the past two years, and led to a sell-off in worldwide tech stocks on Monday.

Asian shares fell on Tuesday following the Wall Street rout, which saw Nvidia drop by 17 percent, losing $600 billion (£482bn) in market value, while the tech-heavy Nasdaq fell 3 percent.

Alphabet, Amazon, Meta, Microsoft and Oracle are planning to spend some $310bn this year on capital expenditures, including AI infrastructure such as data centres and AI chips, according to data from Visible Alpha.

‘Ratchet forward’

Meanwhile, venture capitalists put $100bn into US AI start-ups last year.

DeepSeek’s high-performing open-source models threw questions on how effective such spending will be.

“There’s now an open weight model floating around the internet which you can use to bootstrap any other sufficiently powerful base model into being an AI reasoner,” said Anthropic co-founder Jack Clark in a blog post on Monday, referring to DeepSeek’s technology.

“AI capabilities worldwide just took a one-way ratchet forward,” he wrote, adding, “Kudos to DeepSeek for being so bold as to bring such a change into the world!”

Steve Cohen, founder of $37bn hedge fund Point72, said similarly that DeepSeek’s technology shows that artificial general superintelligence is “coming quick”.

China competition

The unexpected development follows years of trade restrictions by the Trump and Biden administrations designed to impede China’s development of cutting-edge AI.

“We want the US to set the global AI standard, not China,” Meta said in a statement.

OpenAI chief product officer Kevin Weil said on a call with reporters that DeepSeek’s success “underscores how important it is that the US wins this race”.

He added that DeepSeek was “on par with where we were a number of months ago”.

Advertising