Lip-Bu Tan: the Tiger Dad Intel Needs
The most successful tech CEO and VC no one has heard of
Lip-Bu Tan’s first day as Intel’s CEO happens to be during Nvidia’s GTC conference. On that day, he both attended GTC – the flagship annual event of Intel’s rival in AI but also a potential customer for its foundry – and had dinner with one of Intel’s largest customers.
His ability to somehow make “copies” of himself, so he can pack more into a day, turned out to be a daily occurrence during his first 100 days at Intel’s helm. He recently broke his own record of 22 meetings in a single day and would often have two or three dinners on the same night – an “accomplishment” he somewhat proudly, somewhat sheepishly boasted on stage at Cadence’s annual conference a month ago. Perhaps he felt extra comfortable sharing something personal at this event, since he was, after all, Cadence’s CEO before joining Intel.
At 65-years-young and as a grandfather of three grandchildren, Lip-Bu (almost everyone calls him by his given name, so I will call him that here too) is soft-spoken and mild-mannered. His public demeanor is the polar opposite of the loquacious, rah-rah CEO we all expect to see on CNBC or conference stages. Yet, he is likely the most accomplished tech CEO and venture capitalist that no one has heard of.
As a VC at Walden International, the firm he founded in 1987, he has had 145 IPOs ( according to himself). Granted, this number is hard to verify. Also, not every single IPO was a blockbuster listing on the NYSE or NASDAQ; many were on less notable exchanges in Asia. Still, liquidity is liquidity, DPI (Distributed to Paid-In capital) is DPI. Most VC firms could only dream of an IPO count half as many.
At Cadence, the electronic design automation (EDA) juggernaut, Lip-Bu presided over one of the most dramatic turnarounds in tech history. When he took over in 2009, Cadence was trading at $2.42 a share, its previous CEO and four senior VPs all left on the same day, and the company was in total disarray and on the verge of being delisted. When he left the CEO post in December 2021, Cadence was trading at $180-190 a share. The company is now undisputedly one of the big three’s in the EDA world, along with Synopsys and Siemens.
Intel’s current mess is much worse than Cadence’s circa 2009, and a failure to turn things around would not only send shockwaves around the semiconductor industry, but be a stab in the heart of America’s own turnaround to becoming a manufacturing powerhouse again. Without Intel, there would be no Silicon Valley. And Lip-Bu’s energizer-bunny-like superpower and his humble mantra of “underpromise overdeliver” may just be the antidote to Intel’s woe. Unnecessary managerial layers are already shrinking, layoffs are happening, subsidiaries are being divested or spinning out with outside capital, and even a bold write-off of Intel Foundry Service (IFS)’s ill-conceived 18A chipmaking process node is being considered.
A disciplined, tough, but fair “tiger dad”, personified in Lip-Bu Tan, is exactly what Intel needs to survive.
From Johor to Silicon Valley
Born in a port town Johor, Malaysia, Lip-Bu was the youngest of five children. His one older brother and three older sisters all either have PhD degrees and/or are musicians at a concert performance level.
Lip-Bu, on the other hand, could not play any instrument and never got his PhD. His lack of a doctorate degree was something he lamented in shame only half-jokingly well into his forties as a source of constant disappointment from his mother. It’s safe to say that he grew up under classic tiger parenting that would make Amy Chua, who wrote “Battle Hymn of the Tiger Mother”, look soft.
He was, of course, no chump in any academic setting. Lip-Bu breezed through the quantum physics program in three years as an undergraduate at Nanyang University in Singapore. (It was later merged with another institution to become the National University of Singapore.) He then matriculated at MIT. His hope at the time was to get a PhD in nuclear engineering, because he saw energy production as a big problem that he wanted to help solve. He was prescient in identifying both the problem and the potential solution, but the Three Mile Island accident happened at the same time and got in the way. As he recounted that moment in an oral history interview for the Computer Science Museum, after the nuclear meltdown accident, the director of the career placement office at MIT told him point blank to get out of nuclear engineering because nuclear energy has no future in the United States.
In an alternate timeline, had that moment not occurred and Lip-Bu went on to get his PhD in nuclear engineering, then ChatGPT happened exactly as it did, he would likely still be at the center of the AI boom, except working on solving the energy bottleneck. In another less likely but still fascinating alternate timeline, where he had stayed in his hometown and never moved to Singapore nor the US to study, Lip-Bu would still probably play a major role in the AI boom. According to independent analysis, Johor is now one of the densest AI hubs, where Oracle is leading the way in constructing massive AI datacenters, a development no one could have predicted.
Of course, Lip-Bu’s real timeline didn’t turn out that bad. After swiftly dropping out of the PhD program on that placement officer’s advice, he landed in San Francisco, working for a nuclear energy consulting company. The company paid for him to get an MBA at the University of San Francisco at night time, so he could learn finance, accounting, and balance the company’s book as its “CFO” during the day. After completing his MBA, Lip-Bu did a stint at a small UK-based investment bank, Chappell & Company, before striking out on his own as a venture capitalist.
To break into VC, he was hired by Walden Capital. To convince the Walden partners to take a chance on him, he proposed to not take any salary and raise his own fund under the firm’s brand. That first fundraise was a mere $3.3 million, which he cobbled together from his father-in-law and his dad’s old friends in Malaysia. That tiny sum also became the seed that grew into Walden International, the semiconductor-focused venture fund with 100-plus IPOs under its belt.
Walden International started investing almost immediately in Asia after its founding in 1987, at a time when the term “venture capital” might as well be gibberish. Credited with introducing the venture model to the continent, Lip-Bu had to teach and train his local investors on the basics of VC in each Asian country his firm operated in – from investing, to scaling, to exiting. He was instrumental in orchestrating one of the first major IPOs of a Chinese internet company on the NASDAQ: Sina.com in 2000. He also invested in, and was on the board of, the more controversial Semiconductor Manufacturing International Corporation (SMIC), China’s national champion chip fab; he left SMIC’s board in 2019.
His extensive domain expertise in semiconductor and local knowledge of Asia, right as many Asian countries grew to become crucial elements of the global supply chain, made him an attractive board member candidate for many U.S. tech companies. One of the companies that recruited him was Cadence, whose board he joined in 2004, before becoming its CEO in 2009.
Because he was allowed to work two full-time jobs, as a CEO and a VC, the low key Lip-Bu also became a dense node of connections and relationships. Not only did he continue to invest from Walden International while running Cadence, he also co-founded Celesta Capital in 2013, another deep tech focused VC firm. In 2017, an analytics firm that measured the relationship network of tech leaders dubbed him the most connected tech executive that year, beating out more household names, like Sheryl Sandberg and Meg Whitman. No company or notable executive in the semiconductor ecosystem is more than one or two degrees removed from Lip-Bu. They are either a customer, a portfolio company, a partner, or, oftentimes, an acquisition target. He invested in Annapurna Labs, the Israeli chip design startup that Amazon acquired and is now the core team behind AWS’s ASICs product line, Trainium and Inferentia. Habana Labs was also a portfolio company, which Intel acquired to boost its internal chip design capabilities but with much less success.
Throughout this glidepath from a port town in Johor to the center of Silicon Valley, Lip-Bu perhaps only encountered one challenge he has not yet overcome – his English. As the
youngest child, his mother thought it would be a good idea to send her son to a Chinese-speaking school, as opposed to English-speaking schools like all his older siblings, so he could grow up speaking more Chinese and have more Confucian values. Because of this choice, Lip-Bu had a very hard time adjusting to MIT when he first started studying there. His English was halting, broken, and he could not understand the American accent of his professors.
His mother later apologized for this choice, a decision that still affects Lip-Bu’s English today – occasionally broken, but in the most sincere, endearing, grandfatherly way.
Cut or Bust
As endearing as he may come across, Lip-Bu is not your warm and fuzzy CEO. For Intel, things are so dire there is no time for warm and fuzzy. Time is running out for the once beacon of American innovation to slim down, focus on what matters, and operate more like a startup just to stay relevant.
Lip-Bu wasted no time making fixes since taking the helm a few months ago. To help the company focus, he sold off Altera (its FPGA subsidiary) to Silver Lake, a leading tech private equity firm, and shut down its small automotive division. To slim things down, he removed a bunch of managerial layers, expanded his own number of direct reports to drive faster decision-making, and began laying off people in the hundreds and thousands across all functions. He even outsourced most of Intel’s marketing functions to AI (though still managed by the humans at Accenture)! Since Lip-Bu was on Intel’s board for two years from 2022 to 2024, before a rumored clash with Pat Gelsinger, the previous CEO, led to him leaving the board before returning as CEO, he likely already knows all the major problems with Intel. He just needs to fix them, fast!
This may be his first time as Intel’s CEO, but it is not the first time Lip-Bu is tasked with cleaning up an Intel-type mess. When he set out to turn around Cadence in 2009, the mess was
created by long-time Intel executive, Michael Fister, who took the Cadence CEO role in 2004. He brought all the good and bad of Intel with him. One of the major problems that came with the domineering Intel culture was a lack of service mentality. That was fine when you have effectively monopolized your market, which was more or less the case back then with Intel’s dominance of the CPU market. But that posture was toxic when your core product was a service tool meant to cater to the needs of different customers, as was the case for Cadence.
Switching to an engineering-led service mentality would be the painful but necessary culture shift that may determine the fate of Intel’s troubled foundry service, IFS, and by extension, America’s indigenous advanced chipmaking capability. IFS’s problem is not that it isn’t technologically advanced; Intel still possesses bleeding-edge R&D with enough talent to see new ideas come to life. It is also willing to spend the big bucks to buy the best equipment on the market, for example, ASML’s High NA EUV lithography machine, which TSMC has shunned due to cost. But those innovations and investments have always been reserved for Intel itself, not packaged to serve outside customers, who are oftentimes also competitors. This lack of service mentality is a cardinal sin when running a foundry business.
IFS’s awkward identity crisis is on full display in its 18A and 14A manufacturing process. 18A (the 18 represents 1.8nm transistor density) is IFS’s most advanced chipmaking process to date. While it is technologically competitive with TSMC’s equivalent process, it was designed to satisfy Intel’s own next generation CPUs. That means the process did not have all the necessary IP portfolio, partnerships with popular EDA providers like Cadence, and, most importantly, outside customer commitment and buy-in to co-design and co-develop. Any customer would’ve had to make their chips in Intel’s way – zero service mentality. In short, 18A was stuck in the bygone era of “read men have fabs”, a phrase infamously coined by AMD founder Jerry Sanders, while all of Intel’s competitors have moved on to become fabulously fabless.
Intel later realized its foundry service needed to serve more than itself, perhaps because it recognized the complete hollowing out of American advanced chip manufacturing and the dire national security implications, or perhaps because it just needed to qualify for the CHIPS Act subsidies to stay competitive and relevant. In either case, the 18A process was “intercepted” while it was being developed, to add essential elements like PDK (process development kit) to make it palatable to potential outside customers, an unfortunate fact Intel CFO David Zinsner recently admitted.
It was too late. According to most industry insiders, 18A is having a hard time attracting notable customer commitment, with Microsoft, Amazon, and the Department of Defense only ordering small volumes of chips. Almost the entire industry has committed to TSMC’s N2 process node instead (N2 represents 2nm transistor density).
14A (the 14 represents 1.4nm transistor density) is still in its infancy and is being designed with the right service mentality from the get-go. It also has enough innovation to make it technologically attractive, at least for now. One of the key Intel innovations with both 18A and 14A is the backside power delivery network. This technique allows power to be sent more directly to the transistors with less power drop, less heat, and higher transistor density. It gives Intel an edge, but the technology is relatively unproven, and it isn’t something that TSMC cannot replicate once it becomes mature. That’s why even though backside power delivery is being heavily marketed as part of 18A’s offering, it is still struggling.
Although 14A is not expected to reach maturity until 2027, it typically takes two to three years to work side-by-side with customers to co-design, co-optimize, and co-develop advanced chips on a brand new manufacturing process, before tape-out then volume production. That means Lip-Bu needs to sign a marquee customer this year to convince the world that 14A and IFS is viable.
Thus, the rumored news that he may be deprioritizing 18A, leapfrogging to 14A, and doubling-down on making 14A the most advanced and service friendly process node on the market may be the right, if not the only, path forward.
A Tiger Dad’s Tough Love
In a leaked internal Q&A with employees, Lip-Bu told everybody point blank that he did not think Intel was even among the top 10 of semiconductor companies. He asked that everyone “be humble”, because customers are giving Intel a failing grade and this turnaround will be a “marathon”.
These are direct, jarring, but truthful words. In Lip-Bu’s world, an “A” is an “A”, an “F” is an “F”. It is the kind of tough love that only a tiger dad who really cares would give. It is the kind of paranoia that would make Andy Grove, Intel’s third and most iconic CEO, proud.
As Intel’s ninth CEO, Lip-Bu’s brutal honesty is also refreshing when the dominant narrative in America, especially on manufacturing and innovation, is littered with moral victories and “copium”. Meanwhile the rest of the world, particularly Asia, races ahead with better processes, capabilities, and quality of service. Sure, higher tariffs may strong arm some foreign manufacturers to build in America, as TSMC, Foxconn and others have promised to do. But as Intel’s decline has shown, failure in manufacturing comes from undue arrogance and a lack of service mentality. It is a deeply-rooted cultural problem, not a technology problem.
At this point in life, the easiest, and perhaps the most sensible, thing for Lip-Bu to do would be to spend more time with his grandchildren and enjoy the fruits of his lifetime of success. It’s a path his wife wanted him to take even before he took the Intel CEO job.
The second easiest to do would be to take the big job, break up the company in two – a chip design firm and a foundry service – then sell the foundry for spare parts. That would be the typical short-term share price maximization that Wall Street would’ve loved to see. Lip-Bu would make a quick buck, so will existing shareholders, another nice moral victory gets celebrated, while American advanced manufacturing hollows further.
Instead he chose the hard path. Laying people off, divesting periphery businesses, corralling partners, and cashing in all the relationships he has accumulated over forty years in the industry to give IFS a fighting chance, not to mention the grind of 22 meetings per day and 3 dinners per night. The 18A deprioritization rumor, if true, would be the hardest step on this hard path yet. (The day after the rumor was published by Reuters, Intel stock price dropped 4%.)
In the end, none of this may matter. Intel may still continue its inertial decline and eventually fade into oblivion. But at least an honest effort would have been made by the one person who was equipped to make it.
Lip-Bu Tan is the tiger dad that Intel needs. In many ways, he is the tiger dad that much of America needs too.
What amazing timing. Maybe I should try coordinating with Tom Cotton and Fox News to write about someone who's about to earn a "he should be fired" post from the Orange Man.