Tech bosses largely agree the danger DeepSeek poses to OpenAI stays limited for now.
Bloomberg | Bloomberg | Getty Images
The technological advances that Chinese artificial intelligence lab DeepSeek have displayed show the sport is on in relation to U.S.-Sino competition on AI, top tech executives told CNBC.
In a series of interviews at France’s Artificial Intelligence Motion Summit, leaders of several major tech firms told CNBC that the emergence of DeepSeek demonstrates that China cannot be counted out as a serious player in relation to AI innovation.
Last month, DeepSeek shocked global markets with a technical paper saying that one in every of its latest AI models was created with a complete training cost of lower than $6 million — far lower than the billions upon billions of dollars being spent by Big Tech players and Western AI labs reminiscent of OpenAI and Anthropic.
Chris Lehane, chief global affairs officer at OpenAI, told CNBC that DeepSeek’s advanced, low-cost model confirms there’s a “very real competition between U.S.-led, small D democratic AI and CCP [Chinese Communist Party] China-led autocratic, authoritarian AI.”
Many critics of DeepSeek have pointed to apparent censorship by the model in relation to sensitive topics. For instance, when asked in regards to the 1989 Tiananmen Square massacre, DeepSeek’s AI assistant app responds with: “Sorry, that is beyond my current scope. Let’s speak about something else.”
“There’s two countries on the earth that may construct this at scale,” Lehane told CNBC’s Arjun Kharpal on the sidelines of the Paris AI summit Monday. “Imagine if there have been only two countries on the earth that would construct electricity at scale. That is form of how you might have to give it some thought.”
“For us, what DeepSeek really reinforces and reaffirms is that there’s this very real competition with very real stakes,” Lehane added.
Still, tech bosses largely agreed that although DeepSeek’s breakthrough shows China being further along in the worldwide AI race than previously thought, the threat it poses to OpenAI stays limited for now.
‘The sport is on’
DeepSeek says that its latest R1 model, an open-source reasoning model, was capable of rival the performance of OpenAI’s own similar o1 model — only using a less expensive, less energy-intensive process.
That led experts to query the prevailing wisdom within the West of the last several years, which is that China is behind the U.S. on AI development due to export restrictions that make it harder for firms within the country to get their hands on more advanced Nvidia graphics processing units, or GPUs.
GPUs are crucial for training and running AI applications because they excel at parallel processing, meaning they will perform multiple calculations concurrently.
Reid Hoffman, a co-founder of LinkedIn and partner on the enterprise capital firm Greylock Partners, told CNBC Monday that DeepSeek’s latest model is “a giant deal in showing that the sport is on.”
“The competition is afoot with China,” Hoffman said, adding that DeepSeek’s R1 is “a reputable, actionable model.”
Abishur Prakash, founding father of strategic advisory firm The Geopolitical Business, told CNBC that DeepSeek shows the West’s understanding of China stays limited.
“America’s assumed place because the technological captain of the world is not any longer the suitable belief,” Prakash told CNBC in a phone interview.
“That’s the latest establishment now, that the space between the U.S. and China has narrowed almost overnight — but it surely hasn’t narrowed overnight, it has been years of progress,” Prakash said.
“If there’s one takeaway for the West, it’s that their understanding of China is incredibly limited — and we do not know what’s coming next,” he added.
No meaningful threat to U.S. AI — yet
Still, leading AI execs aren’t convinced that DeepSeek poses any form of meaningful risks to the companies of AI labs like OpenAI and Anthropic just yet.
While experts on the entire agree DeepSeek’s AI advances have been impressive, doubts have been raised in regards to the startup’s claims about cost.
A report from semiconductor research firm SemiAnalysis last month estimated that DeepSeek’s hardware expenditure is “well higher” than $500 million over the corporate’s history. DeepSeek was not immediately available for comment when contacted by CNBC.
The report found that DeepSeek’s research and development costs and expenses related to ownership are significant and that generating “synthetic data” for the model to coach on would require “considerable amount of compute.”
Some technologists consider that DeepSeek could have been capable of achieve such a high level of performance by training its models on larger U.S. AI systems.
This method, often known as “distillation,” involves having more powerful AI models evaluate the standard of answers being generated by a more recent model.
It is a claim that OpenAI itself has alluded to, telling CNBC in an announcement last month that it’s reviewing reports that DeepSeek could have “inappropriately” used output data from its models to develop its AI model, a way known as “distillation.”
“Many of the market fear around [DeepSeek] is in truth misplaced,” Hoffman told CNBC. “It still requires large models — it was distilled from large models.”
“I feel the short answer everyone should take is: game on — but large models still really matter,” he added.
Victor Riparbelli, CEO of AI video platform Synthesia, told CNBC that although DeepSeek challenged the “paradigm that brute force scaling is the one method to type of construct higher and higher models,” the concept firms are going to suddenly shift significant amounts of their AI workloads is misguided.
“I still think that once you have a look at users of those technologies, all of the workflows, I feel after we look back in three months’ time, I feel 0.01% of those goes to be moved to Deepseek from OpenAI and Anthropic,” Riparbelli said.
Meredith Whitaker, president of the Signal Foundation, said DeepSeek’s development doesn’t move the needle much for the industry as market momentum continues to be broadly in favor of larger AI models. The Signal Foundation is a nonprofit that supports the encrypted messaging app Signal.
“This shouldn’t be something that is going to disrupt the concentration of power or the geopolitical balance at this stage,” Whitaker told CNBC. “I feel we’ve to maintain our eye on the ball there and recognize that it’s really this ‘greater is best’ paradigm that shouldn’t be reduced through efficiency gains historically, that’s driving this concentration.”
