Fireside Alpha

Great listen: $INTC Lip-Bu Tan x T Rowe's Long View podcast. He's been active on the podcast circuit, and with today's announcement makes sense why. In terms of arc, Tan is a guy who seemingly keeps taking jobs he did not plan to take. He wrote one resume in his life, raised $3.3 million for a fund that now runs more than $4 billion, signed up for a 3-month interim CEO stint at Cadence that became 15 years, and took the Intel seat against the advice of 80% of the people he asked because one friend told him to save the company before he retired and it touched his heart. Some number before the takes: a 5-yr turnaround plan he says he finished in 14 months, a US government stake now up more than 5x, and a CPU-to-GPU ratio he watches moving from 1:8 in training toward 1:1 in inference. 1/ He grew up the youngest of five in Malaysia, the one child his strict mother let off the hook. Tan says he was born in Muar, Johor, just north of Singapore, and followed his mother to Singapore for school before coming to the US for graduate work. His siblings were pushed through piano to the level of performance and composition, and his brother played violin in an orchestra, but Tan says he was too lazy for the piano and spent his time catching spiders and playing basketball and volleyball instead. Being the youngest, he says, came with a lot of freedom, and the music he skipped he now backs from the other side, having served on the board of the San Francisco Opera. 2/ His mother sent him to Chinese school while the other four went to English school, and he still frames his English as a survived second language. Tan says she chose, for some reason he does not fully explain, to send the last child to Chinese school to learn Confucian theory, and that she would apologize to him that his English was not good because it was a second language. He says he survived it. The detail is important to the later story because the same man who frames his English as a handicap is the one who walked into a room with President Trump and talked his way from a resignation demand into a partnership. 3/ He chose physics, then switched to nuclear engineering, partly to escape being measured against his brother. Tan says he started in physics at Nanyang Technological University because he was good at mathematics and physics was good training, and he finished it in three years. But his teachers kept telling him his brother, a Rhodes scholar who went to Cambridge and Oxford and became a cardiologist working on artificial hearts, was better. So he switched fields to do something different, applied to graduate school in the US, was rejected by Berkeley and Stanford, and was accepted at MIT for nuclear engineering, which is the only reason he says he pursued it. 4/ A nuclear accident ended his PhD and a placement officer pushed him into the job market with one company name. Tan says he never finished his PhD because of the Three Mile Island accident, which gutted the prospects for nuclear power plant work. He tells it as comedy: he liked doing homework in the placement office because it had air conditioning, the bored placement director asked his department, heard "nuclear engineering," and told him to get out because nothing was being built. Tan says he wrote exactly one resume in his whole life, to a firm the director named, EDS Nuclear, got flown out for a three-day trip, and took the job. The thread he is drawing is that the defining moves of his career were not planned, they were doors that opened. 5/ He was recruited into his first startup over a no-business lunch at 21 or 22, with nothing to lose. Tan says that during a rotation program at that first company, three executive vice presidents invited him to lunch, and he assumed they wanted to talk budget until they promised no business. The lunch was a pitch to join them in founding a startup. He says he was only 21 or 22 at the time and had nothing to lose, so he went, and that became his first startup experience. He uses it to set up a pattern he repeats across the interview, that the right people pulling you in is more important than the plan you had. 6/ He started Walden by offering to work for free and raising $3.3 million, and built it brick by brick to visit his mother. Tan says that when he turned to venture capital, he knocked on every door and got no offers, so he found a small SBIC firm and told the two partners they did not have to pay him, he would pay for himself. He raised about $3.3 million to start Walden International, which he says now manages more than $4 billion, and across his watch he and his colleagues recorded 153 IPOs and 126 successful M&A exits. Asked his master plan for building out Asia, he says he had none, he just wanted to go visit his mother in Singapore. 7/ Cadence is where he learned to turn a company around, and he became its CEO entirely by accident. Tan says he hesitated to even join the Cadence board because it was the number-two player behind Synopsys, and a friend at Synopsys told him to join the number one instead. He joined number two anyway, then got a call that the CEO had been fired and he was now interim CEO, with the stock at $2.42. He signed up for three months, tried to hand the chairman the key once a search was done, and was told to keep it. He says that three-month stint became 15 years, 12 and a half as CEO and two as executive chairman. As a first-time CEO he read three "CEO for dummies" books cover to cover, opened his inbox to all employees, and says he got more than 3,000 emails a day and promised to answer every one. 8/ One friend's appeal, not the upside, is why he took the Intel job, and 80% of his advisors told him not to. Asked directly why he did it, Tan says about 80% of the CEOs and executives he consulted told him not to take the job, on two grounds: there was no guarantee of success, and he had already proven himself at Cadence, so a failure at Intel would define the rest of his career because people only remember the last one. The thing that moved him was a thoughtful note from a friend and customer in Seattle who wrote that with all his success, Intel is an iconic company critical to the industry and the country, and asked him to save it before he retired. Tan says that touched his heart. He adds that his wife, who had said no when he was a finalist five years earlier, supported him this time because she could see his heart was already in it. 9/ His first act at Intel was structural: pull all of engineering under himself and force bad news to travel fast. Tan says that from day one he had all of engineering report directly to him, the same move he made at Cadence, so he could see for himself where the product, the foundry, and the customer problems actually were. He pairs it with a rule he repeats to the team, that they must bring him the bad news first, because if a customer tells him something his own team withheld, that person is in deep trouble. He frames it as a saying: bad news needs to travel fast, good news can always be celebrated later. Customers, he says, handed him a list of "14 screwed up" mistakes Intel had made, and he wrote down every one and went back to them item by item. 10/ Asked whether Intel's problem was strategy or execution, he answers with culture: too many layers and too many meetings. The host asks directly whether he diagnosed Intel as a strategy problem, an execution problem, or something else. Tan goes at the culture. He says he found management stacked north of 10 layers deep and pushed it down toward five so people would actually be accountable, because in a finger-pointing culture nobody owns the failure. He says more than 100 former top executives told him they spent 89% of their time in meetings, and that the same presentation had to be repeated 36 times before anyone decided anything, often by a decision-maker with no grasp of the product. His fix is what he calls the speed of light: move like a startup, kill meetings with no agenda or deliverable, and read the material himself beforehand so a meeting is five questions and done. 11/ Asked what changes when the US government owns 10% of you, he says it starts with a morning call demanding he resign. The host asks how it affects running the company to have the US government as a large shareholder. Tan answers with the origin: an early-morning phone call telling him the President wanted him to resign over a conflict of interest, which disrupted the gym-and-swim he usually starts the day with. His move was to take himself out of the picture, on the logic that he did not need the job and came to save Intel, then get a meeting with Trump. He says he explained that he was born in Malaysia, grew up in Singapore, went to MIT, and never lived in China, and his children and grandchildren all live in the US, and they listened. The outcome he describes is the government taking 10% against his two-year turnaround pledge, with Trump giving him from Monday to Friday to work out the terms with Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent, on commercial terms with no board seat, a stake he says is now up more than 5x. 12/ Asked where Intel's edge is now, he points to a CPU demand reversal and to building purpose-built silicon across the stack. The host asks where Tan sees his competitive advantage as AMD, Nvidia, and the hyperscalers all push into CPUs. Tan says CPUs became important again with agentic AI and inference, and that the ratio he watches has moved from one CPU to eight GPUs in training toward roughly one-to-one in inference. He concedes the field is crowded, with AMD well respected and Arm-based players including Nvidia moving in. His answer to it is to quietly build inside Intel, hiring top CPU and GPU architects to make purpose-built silicon tuned to different workloads, on the argument that owning CPU, GPU, advanced packaging, and the foundry lets Intel assemble surgical, workload-specific chips no pure-play can. 13/ Asked how he gets back on the foundry treadmill Intel fell off, he sequences it: balance sheet, then talent, then capex. The host frames foundry as a treadmill where the speed jumps every couple of minutes and asks how Intel catches back up. Tan says the first move was strengthening the balance sheet, because the talent he tried to recruit early on told him plainly they would not join a near-bankrupt company. With the balance sheet fixed he could attract leading technologists, force a proud organization to start sharing its data, and send people to physically sit at the weak older-node fabs and not come back until the problem was fixed. He says he raised capex to buy equipment given the long lead times, and ties it back to geography: advanced-technology foundry has only one US player, Intel, while more than 90% of leading-edge manufacturing sits in Taiwan, and the memory crisis is the live reminder of what depending on one geography costs. 14/ The vision is not three Intel businesses, it is one system platform, run like a basketball team. Tan says Intel is described today as a PC business, a data-center business, and a foundry, but the direction is to stitch those into a full system platform, with software, that sells more than a chip. He points to his Foxconn partnership announcement and to his Computex keynote, where he says he deliberately handed parts of the speech to colleagues. He describes his own role as a conductor or a basketball coach, cheering an executive team to deliver rather than doing everything himself, and says the goal is a championship team like the New York Knicks. The point he keeps returning to is that he cannot do it alone, so he is building a deep bench, including potential successors, the same way he did at Cadence. 15/ Asked what would eventually end this capex cycle, he says the brake is applications without ROI, not chips. This is the host's what-breaks-it question, and the highest-signal answer in the interview. Tan says the AI buildout reminds him of the internet but bigger, and that the internet produced a few big winners, an Amazon and a Netflix, who won by being laser-focused on a real application that disrupted an industry. He says he just joined the MIT CEO advisory board, where two professors presented research arguing that companies today do not yet see the ROI or the performance improvement from AI, and that while he may not agree, it keeps him alert. His conclusion is to look past the infrastructure to the application: figure out what the application and the ROI are, then design the silicon down from there, which is how he says he formed Intel's strategy. 16/ He finished the board's five-year plan in 14 months, and reset the clock to ten years on purpose. Tan says the board and he built a five-year plan when he arrived, and at a recent board meeting he was told he had accomplished it in 14 months and needed to aim bigger. So he says he is now drawing up a five- and ten-year plan and telling his team to operate as if he will be there for a decade, so nobody assumes he is leaving. He pairs the longer horizon with succession, saying he will gradually bring in talent who could replace him, the way he did at Cadence. The reset is the candid version of the same instinct elsewhere in the interview, that he raises the target the moment he clears it. 17/ The lightning round is where the operator shows: meditation while swimming, durian, and Star Wars. Asked his one word for Intel today, Tan says technology, and the first technology he loved he gives as quantum. He names Netflix as the non-tech company he most admires, which lines up with the Netflix-and-Amazon framing he used for AI winners. His must-read turns out not to be reading at all but meditation, morning and evening, and he says he does the morning session while swimming. His guilty food is durian, specifically the bitter Mao San Wang or "mountain cat" variety, which he compares to a strong cheese, and his science-fiction pick is Star Wars. Lastly, he keeps making investing calls in the room, including telling Lisa Su to take the AMD job. Tan says that when Lisa Su asked him whether she should take the CEO job at AMD, he told her that her market cap could not end up lower than the company he chaired and had to be higher, and that she went on to do a fabulous job. It is a small aside, but it captures how he treats relationships as the asset, the same instinct that has him recruiting every senior Intel hire himself and crediting his old friend the Seattle customer for the decision that put him in the chair.

SaaS considerations- Robert Smith (Vista Equity) sees enterprise software ending up in 3 states: 1) Agentic state, agentic workflows 2) Enabling delivery of software at lower cost (dev, CS, GTM) 3) No right to exist if no sovereignty of workflows and datasets.