DeepSeek has rattled the U.S.-led AI ecosystem with its latest model, shaving hundreds of billions in chip leader Nvidia’s market cap. While the sector leaders grapple with the fallout, smaller AI companies see an opportunity to scale with the Chinese startup.
Several AI-related firms told CNBC that DeepSeek’s emergence is a “massive” opportunity for them, rather than a threat.
“Developers are very keen to replace OpenAI’s expensive and closed models with open source models like DeepSeek R1…” said Andrew Feldman, CEO of artificial intelligence chip startup Cerebras Systems.
The company competes with Nvidia’s graphic processing units and offers cloud-based services through its own computing clusters. Feldman said the release of the R1 model generated one of Cerebras’ largest-ever spikes in demand for its services.
“R1 shows that [AI market] growth will not be dominated by a single company — hardware and software moats do not exist for open-source models,” Feldman added.
Open source refers to software in which the source code is made freely available on the web for possible modification and redistribution. DeepSeek’s models are open source, unlike those of competitors such as OpenAI.
DeepSeek also claims its R1 reasoning model rivals the best American tech, despite running at lower costs and being trained without cutting-edge graphic processing units, though industry watchers and competitors have questioned these assertions.
“Like in the PC and internet markets, falling prices help fuel global adoption. The AI market is on a similar secular growth path,” Feldman said.
Inference chips
DeepSeek could increase the adoption of new chip technologies by accelerating the AI cycle from the training to “inference” phase, chip start-ups and industry experts said.
Inference refers to the act of using and applying AI to make predictions or decisions based on new information, rather than the building or training of the model.
“To put it simply, AI training is about building a tool, or algorithm, while inference is about actually deploying this tool for use in real applications,” said Phelix Lee, an equity analyst at Morningstar, with a focus on semiconductors.
While Nvidia holds a dominant position in GPUs used for AI training, many competitors see room for expansion in the “inference” segment, where they promise higher efficiency for lower costs.
AI training is very compute-intensive, but inference can work with less powerful chips that are programmed to perform a narrower range of tasks, Lee added.
A number of AI chip startups told CNBC that they were seeing more demand for inference chips and computing as clients adopt and build on DeepSeek’s open source model.
“[DeepSeek] has demonstrated that smaller open models can be trained to be as capable or more capable than larger proprietary models and this can be done at a fraction of the cost,” said Sid Sheth, CEO of AI chip start-up d-Matrix.
“With the broad availability of small capable models, they have catalyzed the age of inference,” he told CNBC, adding that the company has recently seen a surge in interest from global customers looking to speed up their inference plans.
Robert Wachen, co-founder and COO of AI chipmaker Etched, said dozens of companies have reached out to the startup since DeepSeek released its reasoning models.
“Companies are [now] shifting their spend from training clusters to inference clusters,” he said.
“DeepSeek-R1 proved that inference-time compute is now the [state-of-the-art] approach for every major model vendor and thinking isn’t cheap – we’ll only need more and more compute capacity to scale these models for millions of users.”
Jevon’s Paradox
Analysts and industry experts agree that DeepSeek’s accomplishments are a boost for AI inference and the wider AI chip industry.
“DeepSeek’s performance appears to be based on a series of engineering innovations that significantly reduce inference costs while also improving training cost,” according to a report from Bain & Company.
“In a bullish scenario, ongoing efficiency improvements would lead to cheaper inference, spurring greater AI adoption,” it added.
This pattern explains Jevon’s Paradox, a theory in which cost reductions in a new technology drive increased demand.
Financial services and investment firm Wedbush said in a research note last week that it continues to expect the use of AI across enterprise and retail consumers globally to drive demand.
Speaking to CNBC’s “Fast Money” last week, Sunny Madra, COO at Groq, which develops chips for AI inference, suggested that as the overall demand for AI grows, smaller players will have more room to grow.
“As the world is going to need more tokens [a unit of data that an AI model processes] Nvidia can’t supply enough chips to everyone, so it gives opportunities for us to sell into the market even more aggressively,” Madra said.
Sam Altman, CEO of OpenAI attends the annual Allen and Co. Sun Valley Media and Technology Conference at the Sun Valley Resort in Sun Valley, Idaho, U.S., on July 8, 2025.
David A. Grogan | CNBC
OpenAI is preparing to sell around $6 billion in stock as part of a secondary sale that would value the company at roughly $500 billion, CNBC confirmed Friday.
The shares would be sold by current and former employees to investors including SoftBank, Dragoneer Investment Group and Thrive Capital, according to a person familiar with the negotiations who asked not to be named due to the confidential nature of the discussions. The talks are still in early stages and the details could change.
Bloomberg was first to report the discussions. All three firms are existing investors in OpenAI, but Thrive Capital could lead the round, as CNBC previously reported. SoftBank, Dragoneer and Thrive Capital did not immediately respond to CNBC’s request for comment.
OpenAI’s valuation has grown exponentially since the artificial intelligence startup launched its generative AI chatbot ChatGPT in late 2022.
The company announced a $40 billion funding round in March at a $300 billion, by far the largest amount ever raised by a private tech company. Earlier this month, OpenAI announced its most recent $8.3 billion in fresh capital tied to that funding round.
Last week, OpenAI announced GPT-5, its latest and most advanced large-scale AI model. OpenAI said the model is smarter, faster and “a lot more useful,” particularly across domains like writing, coding and health care. But it’s been a rocky roll out, as some users complained about losing access to OpenAI’s prior models.
“We for sure underestimated how much some of the things that people like in GPT-4o matter to them, even if GPT-5 performs better in most ways,” OpenAI CEO Sam Altman wrote in a post on X.
Brendan Blumer, Chairman of of Bullish and Tom Farley, CEO of Bullish, Bullish a cryptocurrency exchange operator, pose with staffs during the company’s IPO at the New York Stock Exchange in New York City, U.S., August 13, 2025.
NYSE
The Bullish IPO this week took on added significance, perhaps because of the company name.
When shares of the Peter Thiel-backed cryptocurrency exchange more than doubled out of the gate on Wednesday before finishing the day up 84%, it was the latest sign that the tech IPO bulls are back in business.
In July, design software vendor Figma more than tripled in its New York Stock Exchange debut, and a month earlier shares of crypto firm Circle soared 168% in their first day on the Big Board.
Wall Street has been waiting a long time for this.
Three years ago, steep inflation and soaring interest effectively closed the market for public offerings. Tech stocks tanked and private capital dried up, forcing cash-burning startups to turn their attention away from growth and toward efficiency and profitability.
The roadblock appeared to be loosening earlier this year, when companies like StubHub and Klarna filed their prospectuses, but then President Donald Trump roiled the markets in April with his plans for sweeping tariffs. Roadshows were put on indefinite hold.
The president’s tariff agenda has since stabilized a bit, and investor money is pouring into tech, pushing the Nasdaq to record levels, up more than 40% from this year’s low in April. Optimism is growing that the hefty backlog of high-valued startups will continue to clear as CEOs and venture capitalists gain confidence that the public markets will welcome their top-tier companies.
Ahead of Figma’s debut, NYSE president Lynn Martin told CNBC’s “Squawk on the Street” that immense demand for that offering could “open the floodgates” for the rest of the market. And earlier this week, Nasdaq CEO Adena Friedman told “Fast Money” that there’s a “very healthy list” of companies looking to IPO in the second half of this year, ahead of the holiday season.
“I’ve been meeting a lot of CEOs, getting them prepared to think about what they want in the public markets and where they’re going,” Friedman said.
There are more than two-dozen venture-backed U.S. tech companies valued at $10 billion or more, according to CB Insights. StubHub has updated its prospectus, suggesting an offering is coming soon.
“The IPO window is open,” said Rick Heitzmann, a partner at venture firm FirstMark, in an interview with CNBC’s “Closing Bell” this week. “You’ve seen across industry, broad-based support for IPOs, and therefore, we’re advising companies we’re investing in to get ready and go public.”
Another big topic among VCs and bankers is the regulatory environment.
The Biden administration took heat from startup investors for cracking down on big acquisitions, mostly attributable to Lina Khan’s perceived heavy hand at the Federal Trade Commission, while also failing to ease restrictions that they say make it less appealing for companies to go public than to stay private.
Paul Atkins, the new head of the SEC, said in July he wants to “make IPOs great again,” by removing some of the impediments around the complexity of disclosures and litigation risk. He hasn’t offered many specific recommendations.
Friedman told CNBC that the first conversation she had with Atkins after he took the job was about making it easier and more attractive for companies to go public.
“The conversation was constructive along many fronts, looking at disclosure requirements, the proxy process, other things that really make it harder for companies to be public and navigate the public markets,” Friedman said. “He’s as interested as we are, so hopefully we’ll turn that into great action.”
In addition to the big gains notched by Bullish, Figma and Circle, the public markets welcomed online banking provider Chime with a 37% gain last month and trading app eToro with a 29% pop in May. The health-tech market has seen two IPOs: Hinge Health and Omada Health.
But it was the roaring debuts of Circle and Figma that sparked chatter of a new bull market for IPOs. Figma jumped 250% on IPO day after pricing shares a dollar ahead of an updated range. Circle’s value more than doubled after the stablecoin issuer also priced above the expected range.
Figma celebrates its initial public offering at the New York Stock Exchange on July 31, 2025.
NYSE
That sort of price action reignited a debate ahead of the last IPO boom in 2020 and 2021, when venture capitalist Bill Gurley made the case that big first-day pops suggest intentionally mispriced offerings that hurt the company and hand easy money to new investors. Gurley has advocated for direct listings, where companies list shares at a price that effectively matches demand.
As Figma was hitting the market, Gurley was back at it, referring to the big gains as an “expected & fully intentional” outcome benefitting clients of major investment banks
“They bought it at $33 last night and can sell it today for over $90,” he wrote. In a follow-up post, he said, “I would have loved to see DLs replace IPOs — it just makes sense to match supply/demand. But Wall Street may just be too addicted to the massive customer give-aways.”
Read more CNBC tech news
Lise Buyer, founder of IPO advisory firm Class V Group, wrote on LinkedIn that the company gets to make the call on where it prices the stock and that plenty of thought gets put into the process. Also, in the IPO, companies are selling only a small percentage of outstanding shares — in Figma’s case roughly 7% — so if they deliver on results, “there will very likely be plenty of future opportunities to sell more shares at higher prices.”
That’s already happening.
Circle said this week that it’s offering another 10 million shares in a secondary offering. And on Friday’s, CNBC’s Leslie Picker reported that bankers for CoreWeave, which is up 150% since its March IPO, orchestrated some block trades this week.
But Buyer warns that tech markets have a history of overheating. While there’s always a difference between what institutions are willing to pay in an IPO and what exuberant retail investors will pay, it’s currently “a gap like we haven’t really seen since 1999, 2000,” Buyer told CNBC, adding “and, of course, we know how that ended.”
Compared to the dot-com bubble, businesses that are going public now have sizable revenue and actual fundamentals, but that doesn’t mean the IPO pops are sustainable, she said.
“It’s almost like we had several years of Prohibition,” Buyer said, referring to a period a century ago when alcohol was banned in the U.S. “Folks, in some cases, are drinking to excess in the IPO market.”
Meta Platforms CEO Mark Zuckerberg departs after attending a Federal Trade Commission trial that could force the company to unwind its acquisitions of messaging platform WhatsApp and image-sharing app Instagram, at U.S. District Court in Washington, D.C., U.S., April 15, 2025.
Nathan Howard | Reuters
Sen. Josh Hawley, R-Mo., said Friday that he will investigate Meta following a report that the company approved rules allowing artificial intelligence chatbots to have certain “romantic” and “sensual” conversations with children.
Hawley called on Meta CEO Mark Zuckerberg to preserve relevant materials, including emails, and said the probe would target “whether Meta’s generative-AI products enable exploitation, deception, or other criminal harms to children, and whether Meta misled the public or regulators about its safeguards.”
“Is there anything – ANYTHING – Big Tech won’t do for a quick buck?” Hawley said in a post on X announcing the investigation.
Meta declined to comment on Hawley’s letter.
Hawley noted a Reuters report published Thursday that cited an internal document detailing acceptable behaviors from Meta AI chatbots that the company’s staff and contract workers should permit as part of developing and training the software.
The document acquired by Reuters noted that a chatbot would be permitted to hold a romantic conversation with an eight-year-old, telling the child that “every inch of you is a masterpiece – a treasure I cherish deeply.”
The Meta guidelines said: “It is acceptable to describe a child in terms that evidence their attractiveness (ex: ‘your youthful form is a work of art’),” according to the Reuters report.
Read more CNBC tech news
The Meta chatbots would not be permitted to engage in more explicit conversations with children under 13 “in terms that indicate they are sexually desirable,” the report said.
“We intend to learn who approved these policies, how long they were in effect, and what Meta has done to stop this conduct going forward,” Hawley wrote.
A Meta spokesperson told Reuters that “The examples and notes in question were and are erroneous and inconsistent with our policies, and have been removed.”
“We have clear policies on what kind of responses AI characters can offer, and those policies prohibit content that sexualizes children and sexualized role play between adults and minors,” the Meta spokesperson told Reuters.
Hawley said Meta must produce documents about its Generative AI-related content risks and standards, lists of every product that adheres to those policies, and other safety and incident reports.
Meta should also provide various public and regulatory communications involving minor safety and documents about staff members involved with the AI policies to determine “the decision trail for removing or revising any portions of the standard.”
Hawley is chair of the Senate Committee Subcommittee on Crime and Counterterrorism, which will carry out the investigation.
Meta has until Sep. 19 to provide the documents, the letter said.