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.
Google CEO Sundar Pichai addresses the crowd during Google’s annual I/O developers conference in Mountain View, California on May 20, 2025.
Camille Cohen | AFP | Getty Images
The Google Doodle is Alphabet’s most valuable piece of real estate, and on Tuesday, the company used that space to promote “AI Mode,” its latest AI search product.
Google’s Chrome browser landing pages and Google’s home page featured an animated image that, when clicked, leads users to AI Mode, the company’s latest search product. The doodle image also includes a share button.
The promotion of AI Mode on the Google Doodle comes as the tech company makes efforts to expose more users to its latest AI features amid pressure from artificial intelligence startups. That includes OpenAI which makes ChatGPT, Anthropic which makes Claude and Perplexity AI, which bills itself as an “AI-powered answer engine.”
Google’s “Doodle” Tuesday directed users to its search chatbot-like experience “AI Mode”
AI Mode is Google’s chatbot-like experience for complex user questions. The company began displaying AI Mode alongside its search results page in March.
“Search whatever’s on your mind and get AI-powered responses,” the product description reads when clicked from the home page.
AI Mode is powered by Google’s flagship AI model Gemini, and the tool has rolled out to more U.S. users since its launch. Users can ask AI Mode questions using text, voice or images. Google says AI Mode makes it easier to find answers to complex questions that might have previously required multiple searches.
In May, Google tested the AI Mode feature directly beneath the Google search bar, replacing the “I’m Feeling Lucky” widget — a place where Google rarely makes changes.
Disposable diapers are a massive environmental offender. Roughly 300,000 of them are sent to landfills or incinerated every minute, according to the World Economic Forum, and they take hundreds of years to decompose. It’s a $60 billion business.
One alternative approach has been compostable diapers, which can be made out of wood pulp or bamboo. But composting services aren’t universally available and some of the products are less absorbent than normal nappies, critics say.
A growing number of parents are also turning to cloth diapers, but they only make up about 20% of the U.S. market.
ZymoChem is attacking the diaper problem from a different angle. Harshal Chokhawala, CEO of ZymoChem, said that 60% to 80% of a typical diaper consists of fossil-based plastics. And half of that is an ingredient called super absorbent polymer, or SAP.
“What we have created is a low carbon footprint bio-based and biodegradable version of this super absorbent polymer,” Chokhawala said.
ZymoChem, with operations in San Leandro, California, and Burlington, Vermont, invented this new type of absorbent by using a fermentation process to convert a renewable resource — sugar — from corn into biodegradable materials. It’s similar to making beer.
“We’re at a point now where we’re very close to being at cost parity with fossil based manufacturing of super absorbents,” said Chokhawala.
The company’s drop-in absorbents can be added into other diapers, which makes it different from environmentally conscious companies like Charlie Banana, Kudos and Hiro, which sell their own brand of diapers.
ZymoChem doesn’t yet have a diaper product on the market. But Lindy Fishburne, managing partner at Breakout Ventures and an investor in the company, says it’s a scalable model.
“Being able to build and grow with biology allows us to unlock a circular economy and a supply chain that is no longer petro-derived, which opens up the opportunities of where you can manufacture and how you secure supply chains,” Fishburne said.
Other investors include Toyota Ventures, GS Futures, KDT Ventures, Cavallo Ventures and Lululemon. The company has raised a total of $35 million.
The Lululemon partnership shows that it’s not just about diapers. ZymoChem’s bio-based materials can also be used in other hygiene products and in bio-based nylon. Lululemon recently said it will use it in some of its leggings, which were traditionally made with petroleum.
Dylan Field, co-founder and CEO of Figma, appears at the Bloomberg Technology Summit in San Francisco on May 9, 2024.
David Paul Morris | Bloomberg | Getty Images
Design software company Figma filed for an IPO on Tuesday, and plans to trade on the New York Stock Exchange under ticker symbol “FIG.”
The offering would be one of the hotly anticipated IPOs in recent years given Figma’s growth rate and its high private market valuation. In late 2023, a $20 billion acquisition agreement with Adobe was scrapped due to regulatory concerns in the U.K. That led Adobe to pay Figma a $1 billion termination fee.
Revenue in the first quarter increased 46% to $228.2 million from $156.2 million in the same period a year ago, according to Figma’s prospectus. The company recorded a net income of $44.9 million, compared to $13.5 million a year earlier.
As of March 31, Figma had around 450,000 customers. Of those, 1,031 were contributing at least $100,000 a year to annual revenue, up 47% from a year earlier. Clients include Amazon Web Services, Google, Microsoft and Netflix. More than half of revenue comes from outside the U.S.
Figma didn’t say how many shares it plans to sell in the IPO. The company was valued at $12.5 billion in a tender offer last year, and in April it announced that it had confidentially filed for an IPO with the SEC.
Wall Street banks predicted a rush of IPOs after Donald Trump won the U.S. presidential election in November following a dry spell dating back to late 2021, when soaring inflation and rising interest rates pushed investors out of risky assets. While President Trump’s announcement of sweeping tariffs in April roiled markets and led a number of companies to delay their plans, activity has been picking up of late.
Stablecoin issuer Circle doubled in value in its early June debut and is now up more than sixfold from its IPO price for a market cap of almost $43 billion. Online banking company Chime also debuted in June, following Hinge Health’s IPO in May. Artificial infrastructure provider CoreWeave, which went public in March, jumped 46% in June and has quadrupled since its offering.
Buy now, pay later company Klarna, based in the U.K., filed for a U.S. IPO in March, as did ticket marketplace StubHub.
Figma was founded in 2012 by CEO Dylan Field, 33, and Evan Wallace, and is based in San Francisco. The company had 1,646 employees as of March 31.
Before establishing Figma, Field spent over two years at Brown University, where he met Wallace. Field then took a Thiel Fellowship “to pursue entrepreneurial projects,” according to the filing. The two-year program that Founders Fund partner Peter Thiel established in 2011 gives young entrepreneurs a $200,000 grant along with support from founders and investors, according to an online description.
Field is the biggest individual owner of Figma, with 56.6 million Class B shares and 51.1% of voting power ahead of the IPO. He said in a letter to investors that it was time for Figma to buck the “trend of many amazing companies staying privately indefinitely.”
Databricks, SpaceX and Stripe are among high-valued companies that are still private.
“Some of the obvious benefits such as good corporate hygiene, brand awareness, liquidity, stronger currency and access to capital markets apply,” he wrote, explaining the decision. “More importantly, I like the idea of our community sharing in the ownership of Figma — and the best way to accomplish this is through public markets.”
Field added that as a public company, investors should “expect us to take big swings,” including through acquisitions. In April Figma bought the assets and team of an unnamed technology company for $14 million, according to the filing.
The IPO will also mark another much-needed win for Silicon Valley venture firms, which are in need of returns after the multi-year slump. Index Ventures is the largest outside shareholder, with a 17% stake before the offering, according to the filing. Greylock owns 16%, Kleiner Perkins controls 14% and Sequoia has a stake of 8.7%.
Figma said it faces “intense competition” and that loss of market share would “adversely affect our business,” but didn’t name any specific competitors.
Over 13 million people use Figma per month, and only one-third of them are designers, according to the filing. In March the company announced Figma Sites, a tool that turns designs into working websites. It’s one of a few new products that diversify the company away from its collaborative service for crafting app and website designs.
As of March 31, Figma had $1.54 billion in cash, cash equivalents and marketable securities.
Using its cash, Figma has begun investing in digital currencies. In 2024, Figma’s board authorized a $55 million investment into a Bitwise Bitcoin exchange-traded fund. As of March 31, the holding was worth $69.5 million, according to the filing. In May, the board approved a $30 million investment in Bitcoin, and Figma spent the money on USD Coin, which is a stablecoin.