Intel’s CEO Lip-Bu Tan speaks at the company’s Annual Manufacturing Technology Conference in San Jose, California, U.S. April 29, 2025.
Laure Andrillon | Reuters
When Lip-Bu Tan was named CEO of Intel a little over two months ago, he brought with him plenty of name recognition. Tan spent 12 years running Cadence Design Systems and before that was a prominent venture capitalist. He’s also held board seats at SoftBank and Hewlett Packard Enterprise.
“Lip-Bu’s Rolodex is like nobody else’s in the semiconductor industry,” Intel CFO David Zinsner said at a financial conference this month. Zinsner said Tan recently met with 22 potential customers and partners in a single day.
At age 65, Tan is going to need more than a vast database of contacts and four decades of operating and investing experience to turn around the company that put the silicon in Silicon Valley but is struggling to stay relevant in a market that’s increasingly centered around artificial intelligence.
Once the world’s largest chipmaker, Intel has lost 70% of its value since early 2020. It’s roughly flat since Tan was named as CEO on March 12.
Tan’s jam-packed schedule in large part reflects a need to change the industry’s perception of Intel. No longer the dominant player in semiconductors, Intel is trying to pivot into chip manufacturing, especially as the U.S. focuses on investing in onshoring critical technologies. Tan has been listening to customers to find out specific technical requirements they would need from Intel as a foundry, he’s said in public remarks.
Under Tan’s predecessor, Pat Gelsinger, Intel spent $90 billion between 2021 and 2024 on building the company’s foundry operations and unlocking additional U.S. government funding. Capital expenditures in 2025 are expected to reach $18 billion.
Investors, and eventually the board, lost trust in Gelsinger’s ability to generate much of a return on that investment, leading to his ouster late last year. In an industry where roadmaps and capital plans are measured in five-year increments, Tan is under pressure to start building confidence immediately.
“The foundry business, it operates at a different time scale,” said Alvin Nguyen, an analyst at Forrester. “It operates with a level of investment that is tough to stomach, and very few publicly traded companies can deal with it.”
Intel faces a plethora of other challenges that all predate Tan’s tenure. The company’s central processors, or CPUs, that for decades were the most expensive and important part in computers, have been supplanted by AI chips, primarily graphics processing units, or GPUs, from Nvidia. Meanwhile, Advanced Micro Devices has picked up substantial market share in CPUs and server chips, and Qualcomm has emerged as a big challenger as well.
Tan is working on an AI strategy under Sachin Katti, who was named chief technology officer in April after joining the company in 2021.
Tan was born in Malaysia and raised in Singapore. He moved to the U.S. in the 1970s and studied nuclear engineering at the Massachusetts Institute of Technology. He’s since touched just about every aspect of the chip industry.
Before joining Intel, he was CEO of Cadence, which makes electronic design automation, or EDA, software, widely used by engineers at fabless chip companies to design new processors. As a venture capitalist at Walden International, Tan invested in Semiconductor Manufacturing International Corporation, China’s national foundry, in 2001, and was on the board for over a decade.
He’s now betting on Intel, not just with his time but also his wallet. When he became CEO, he bought $25 million of shares, which he’ll have to hold in order to earn his full compensation over the next five years.
Tan has been keeping a fairly low profile since starting the gig in March. He’s yet to sit for a press interview, and Intel declined to make Tan available for this story. But in his two public speeches as CEO at Intel events, he’s laid out elements of his strategy.
“We need to do a better job — make it easier for all of you to use our technology,” Tan said at a foundry event earlier this month. “We will rapidly embrace industrial standards, EDA tools and best design practices.”
One big customer
The fastest way to change the trajectory would be to announce a big foundry customer. Locking in substantial orders would serve as both a vote of approval to other potential customers and a signal to Wall Street that all those expenses will soon start turning into revenue.
“One Nvidia, one Qualcomm, one Apple, one something of volume that really shows this meaningful commitment for the fab to build significant volume would really change the whole narrative,” said Daniel Newman, CEO of industry research firm The Futurum Group.
Tan’s second public appearance as CEO came in April at Intel’s Foundry Direct Connect event in San Jose, California, a few miles from the company’s headquarters. There he hinted at one of his key objectives: rebuilding confidence.
“This is a truly a service business, and that is built on the foundational principle of trust,” Tan said. “You have to be patient to earn your trust.”
Intel wafers are displayed on stage at the company’s Annual Manufacturing Technology Conference in San Jose, California, U.S. April 29, 2025.
Laure Andrillon | Reuters
At the event, populated largely by people from the insular world of chip design and manufacturing, Tan directly addressed foundry customers, discussing the company’s specific technologies in power and packaging that put it in position to take on Taiwan Semiconductor Manufacturing Company, the largest foundry in the world.
Outside the convention center, banners still hung promoting the Nvidia GTC conference, which had taken place the prior month and packed the building’s ballroom.
Tan mostly acted like an emcee, calling up the CEOs of chip design partners such as Synopsys, Cadence and Siemens, who took the stage to discuss using Intel’s technology.
A key issue for Intel to address is the broadening of its foundry, which was originally designed for its own chip design teams, meaning some of the tools and infrastructure were company-specific. Intel has given the name 18A to its chip technology that it hopes to start producing in volume this year.
“One thing about 18A was, it was developed initially as just something for Intel, and we intercepted it relatively early,” Zinsner said earlier this month. That allowed the company to develop process design kits, or PDKs, “for the industry, but it still was not from the ground up developed as a foundry node,” he said.
Zinsner said the company’s next chip generation, 14A, will be built for external customers. Analysts say that 18A may be Intel’s first foundry process that could beat TSMC’s rival process to market.
Tan also recognizes that TSMC has created an industry standard, so using the same tools and technology would allow companies to more simply bring over work from other foundries. He said Intel is making its PDK easier to use.
“My top priority is to make it easier for the ecosystem to do business with Intel,” he said.
One of the speakers at the event was Anirudh Devgan, who succeeded Tan as CEO of Cadence. Tan asked Devgan what AI chip companies need to see if they’re to build on Intel. Devgan said the most important consideration is the need to focus on what the customer wants rather than what Intel prefers.
“Intel Foundry, as you all know, is like the service business, so the customer comes first,” Devgan said. “I know Lip-Bu has very good instincts to understand what the customer wants.”
It’s a stark change in approach for a company that for decades was focused on selling its own chips and not on creating an ecosystem. In a podcast earlier this year, TSMC founder Morris Chang said that Intel, during its glory years, acted “like they were the only guy with microprocessors.”
If there was a disappointment at the Intel event, it was the lack of an announcement about a major new customer.
Zinsner previously said, in response to a question about how many customers Intel had signed up for its foundry, that the company first needs to “eat its own dogfood,” indicating that the 18A process would be primarily used by Intel itself.
Leaner company
While Tan looks outward for business development, he’s turning inward to try to fix corporate culture, flattening the organization, which grew fiercely in recent years as it staffed up to build the foundry unit.
Intel said on its April earnings call that job cuts will come this quarter, though the company didn’t provide a specific number. An Intel representative declined to comment on the matter. Intel announced in August, while Gelsinger was still in charge, that it was laying off 15,000 employees and would explore cuts in its portfolio.
Wall Street welcomes more belt tightening but warns that the company can’t cut its way to a successful revival.
Deutsche Bank’s Ross Seymour, who recommends holding the stock, wrote in a May note that, even with the “welcome and necessary cost-cutting actions,” the company’s “path to meaningful earnings/free cash flow generation remains cloudy and highly dependent on a turnaround” in the foundry business.
Equally important to Tan is getting rid of what he views as too much bureaucracy.
“It has been eye-opening for me to see how much time and energy is spent on internal administrative work that does not move our business forward,” Tan wrote, in a memo to employees in April.
He said Intel would have to learn how to do more with fewer people and that employees must be back in the office for at least four days a week by September.
“I’ve been surprised to learn that, in recent years, the most important KPI for many managers at Intel has been the size of their teams,” Tan wrote, referring to key performance indicators. “Going forward, this will not be the case.”
Tan also promoted several engineering leaders, giving him greater visibility into the organization. Zinsner said Tan has between 15 and 17 direct reports, because he wants to be closer to the “lowest” levels of the organization.
“He’s hearing the good, the bad, the ugly of what’s going on, so that he can help address those,” Zinsner said.
Nvidia CEO Jensen Huang in Taipei, Taiwan, on June 2, 2024.
Ann Wang | Reuters
Nvidia’s Blackwell Ultra chips, the company’s next-generation graphics processor for artificial intelligence, have been commercially deployed at CoreWeave, the companies announced on Thursday.
CoreWeave has received shipments of Dell-built shipments based around Nvidia’s GB300 NVL72 AI systems, Dell said on Thursday. It’s the first cloud provider to install systems based around Blackwell Ultra.
The Blackwell Ultra is Nvidia’s latest chip, expected to ship in volume during the rest of the year. The systems that CoreWeave is installing are liquid-cooled and include 72 Blackwell Ultra GPUs and 36 Nvidia Grace CPUs. The systems are assembled and tested in the U.S., Dell said.
CoreWeave shares rose 6% during trading on Thursday, Dell shares were up about 2% and Nvidia rose less than 2%.
The announcement is a milestone for Nvidia.
Read more CNBC tech news
AI developers still clamor for the latest Nvidia chips, which have improvements that make them better for training and deploying models.
Nvidia said Blackwell Ultra can produce 50 times more AI content than its predecessor, Blackwell.
Investors closely watch how Nvidia manages the transition when it announces new AI chips to see if there are production issues or delays. Nvidia CFO Colette Kress said in May that Blackwell Ultra shipments would start in the current quarter.
It’s also a win for CoreWeave, a cloud provider that rents access to Nvidia GPUs to other clouds and AI developers. Although CoreWeave is smaller than the cloud services operated by Amazon, Google, and Microsoft, its ability to offer Nvidia’s latest chips first give it a way to differentiate itself.
CoreWeave historically has a close relationship with Nvidia, which owns a stake in the cloud provider. CoreWeave went public earlier this year, and the stock price has quadrupled since its IPO.
Jeremy Allaire, CEO and co-founder of Circle Internet Group, the issuer of one of the world’s biggest stablecoins, and Circle Internet Group co-founder Sean Neville react as they ring the opening bell, on the day of the company’s IPO, in New York City, U.S., June 5, 2025.
NYSE
For over three years, venture capital firms have been waiting for this moment.
Tech IPOs came to a virtual standstill in early 2022 due to soaring inflation and rising interest rates, while big acquisitions were mostly off the table as increased regulatory scrutiny in the U.S. and Europe turned away potential buyers.
Though it’s too soon to say those days are entirely in the past, the first half of 2025 showed signs of momentum, with June in particular producing much-needed returns for Silicon Valley’s startup financiers. In all, there were five tech IPOs last month, accelerating from a monthly average of two since January, according to data from CB Insights.
Highlighting that group was crypto company Circle, which more than doubled in its New York Stock Exchange debut on June 5, and is now up sixfold from its IPO price for a market cap of $42 billion. The stock got a big boost in mid-June after the Senate passed the GENIUS Act, which would establish a federal framework for U.S. dollar-pegged stablecoins.
Venture firms General Catalyst, Breyer Capital and Accel now own a combined $8 billion worth of Circle stock even after selling a fraction of their holdings in the offering. Silicon Valley stalwarts Greylock, Kleiner Perkins and Sequoia Capital are set to soon profit from Figma’s IPO, after the design software vendor filed its public prospectus on Tuesday. Since its $20 billion acquisition agreement with Adobe was scrapped in late 2023, Figma has been one of the most hotly anticipated IPOs in startup land.
It’s “refreshing and something that we’ve been waiting for for a long time,” said Eric Hippeau, managing partner at early-stage venture firm Lerer Hippeau, regarding the exit environment. “I’m not sure that we are confident that this can be a sustained trend yet, but it’s been very encouraging.”
Another positive sign for the industry the past couple months was the performance of artificial infrastructure provider CoreWeave, which went public in late March. The stock was relatively stagnant for its first month on the market but shot up 170% in May and another 47% in June.
For venture firms, long considered the lifeblood of risky tech startups, IPOs are essential in order to generate profits for the university endowments, foundations and pension funds that allocate a portion of their capital to the asset class. Without handsome returns, there’s little incentive for limited partners to put money into future funds.
After a record year in 2021, which saw 155 U.S. venture-backed IPOs raise $60.4 billion, according to data from University of Florida finance professor Jay Ritter, every year since has been relatively dismal. There were 13 such offerings in 2022, followed by 18 in 2023 and 30 last year, collectively raising $13.3 billion, Ritter’s data shows.
The slowdown followed the Federal Reserve’s aggressive rate-hiking campaign in 2022, meant to slow crippling inflation. As the lower-growth environment extended into years two and three, venture firms faced increasing pressure to return cash to investors.
‘Backlog of liquidity’
In its 2024 yearbook, the National Venture Capital Association said that even with a 34% increase in U.S. VC exit value last year to $98 billion, that number is 87% below the 2021 peak and less than half the average for the four years from 2017 through 2020. It’s a troubling dynamic for the 58,000 venture-backed companies that have raised a total of $947 billion from investors, according to the annual report, which is produced by the NVCA and PitchBook.
“This backlog of liquidity drought risks creating a ‘zombie company’ cohort — businesses generating operational cash flow but lacking credible exit prospects,” the report said.
Other than Circle, the latest crop of IPOs mostly consists of smaller and lesser-known brands. Health-tech companies Hinge Health and Omada Health are valued at about $3.5 billion and $1 billion, respectively. Etoro, an online trading platform, has a market cap of just over $5 billion. Online banking provider Chime Financial has a higher profile due largely to a years-long marketing blitz and is valued at close to $11.5 billion.
Meanwhile, the highest valued private companies like SpaceX, Stripe and Databricks remain on the sidelines, and AI highfliers OpenAI and Anthropic continue to raise massive amounts of cash with no intention of going public anytime soon.
Still, venture capitalists told CNBC that there are plenty of companies with the financial metrics to be public, and that more of them are readying for the process.
“The IPO market is starting to open and the VC world is cautiously optimistic,” said Rick Heitzmann, a partner at venture firm FirstMark in New York. “We are preparing companies for the next wave of public offerings.”
There are other ways to make money in the meantime. Secondary sales, a process that involves selling private shares to new investors, are on the rise, allowing early employees and investors to get some liquidity.
And then there’s what Mark Zuckerberg is doing, as he tries to position his company at the center of AI innovation and development.
Mark Zuckerberg, chief executive officer of Meta Platforms Inc., during the Meta Connect event on Wednesday, Sept. 25, 2024.
Bloomberg | Bloomberg | Getty Images
Last month, Meta announced a $14 billion bet on Scale AI, taking a 49% stake in the AI startup in exchange for poaching founder Alexandr Wang and a small group of his top engineers. The deal effectively bought out half of the stock owned by investors, leaving them with the opportunity to make money on the rest of their holdings, should a future acquisition or IPO take place.
The deal is a big win for Accel, which led Scale AI’s Series A round in 2017, and is poised to earn more than $2.5 billion in the transaction. Index Ventures led the Series B in 2018, and Peter Thiel’s Founders Fund led the Series C the following year at a valuation of over $1 billion.
Investors now hope the Federal Reserve will move toward a rate-cutting campaign, though the central bank hasn’t committed to one. There’s also ongoing optimism that regulators will make going public less burdensome. Last week, Reuters reported, citing sources familiar with the matter, that U.S. stock exchanges and the SEC have discussed loosening regulations to make IPOs more enticing.
Mike Bellin, who heads consulting firm PwC’s U.S. IPO practice, said he anticipates a diversity of IPOs across sectors in the second half of the year. According to data from PwC, pharma and fintech were among the most active sectors for deals through the end of May.
While the recent trend in IPO activity is an encouraging sign for investors, potential roadblocks remain.
Tariffs and geopolitical uncertainty delayed IPO plans from companies including Klarna and StubHub in April. Neither has provided an update on when they plan to debut.
FirstMark’s Heitzmann said the path forward is “not at all clear,” adding that he wants to see a strong quarter of economic stability and growth before confidently saying that the market is wide open.
Additionally, other than CoreWeave and Circle, recent tech IPOs haven’t had big pops. Hinge Health, Chime and eToro have seen relatively modest gains from their offer price, while Omada Health is down.
But virtually any activity beats what VCs were experiencing the last few years. Overall, Hippeau said recent IPO trends are generally encouraging.
“There’s starting to be kind of light at the end of the tunnel,” Hippeau said.
The position was valued at about $160 million as of Wednesday’s close.
Tripadvisor shares have been flat since the start of the year after plummeting more than 30% in 2024. Last year, the travel review and booking company said it created a special committee to explore potential options.
Read more CNBC tech news
Starboard Value has gained a reputation for pushing for changes such as new CEOs and cost cuts by acquiring significant shares in companies.