Lip-Bu Tan's Year: Intel's New CEO Faces a Foundry Breakup Question, a CHIPS-Act Knife-Edge, and a Washington Skeptic

On 13 June 2026, three months after Pat Gelsinger was pushed out, Intel's board installed Lip-Bu Tan as CEO. The market delivered its verdict within hours: shares rose roughly 14% on the day and around 18% over the trailing week, rewarding the simple fact that someone credible was sitting in the chair. The harder verdict is still ahead, and the clock on it is short. Insiders, board members and bankers who have known Tan since the 1990s describe a window of about a year before the next inflection point — the moment when the board, the White House, and the foundry customers will each have rendered a judgment on whether the new boss has bought Intel another cycle or merely postponed a breakup.

The thesis that emerges from the TBPN coverage of the hire, and from the broader market read that surrounds it, is that Tan has been handed a binary. Either Intel executes a clean separation of its design and its foundry businesses — the option Wall Street has spent twelve months quietly begging for — or it commits to the multi-year, multi-tens-of-billions capital programme required to make its own fabs credible in the leading-edge era. The third path, the one Intel has been stumbling along since 2024, is no longer available: trying to subsidise a foundry buildout out of a shrinking CPU cash cow while the rest of the industry converts to AI accelerators and custom silicon.

The preconditions for the split are visible. Intel has shed roughly 15,000 jobs in the past year, about a quarter of its workforce, and a culture reset under a new CEO almost always means more headcount action. Two-thirds of the company's market value has evaporated over the last four years — the AI boom happened, and Intel was largely elsewhere. Bankers have tested the waters: a deal to sell all or part of Intel to Broadcom, Qualcomm, GlobalFoundries, or TSMC was floated in early-stage discussions and has been quietly taken off the table for the immediate future. Tan's job, in this reading, is to revive the auction through execution rather than to consummate it on day one.

A CEO with a 3,000% track record and a Walden problem

Tan's professional reputation is built on a thirteen-year run at Cadence Design Systems, where the share price returned more than 3,000% on his watch. That is the credential the board is buying, and it is the credential a sceptical market is willing to underwrite for at least one earnings cycle. The complication is biographical. Before Cadence, Tan ran Walden International, a Silicon Valley venture firm that was an early institutional investor across Asia. Walden was an early backer of SMIC — Semiconductor Manufacturing International Corporation, China's leading domestic foundry, which the US Commerce Department placed on the entity list in 2020 over alleged ties to the Chinese military. Walden sold its last SMIC stake in 2021, but the position is on the record. In 2023, the House Select Committee on the CCP sent Tan a letter flagging hundreds of millions of dollars in Walden investments that flowed to Chinese firms with alleged military or human-rights exposure.

For most of the last two decades, that history was a footnote in Tan's deal-maker bio. In 2026, with the Trump administration signalling intent to unwind the CHIPS Act, it is a political variable. The CHIPS framework allocated more than $50 billion to US semiconductor capacity, of which Intel stood to receive up to $8 billion contingent on fab development milestones. The administration's posture is that those milestones have slipped; Intel's fab buildouts have faced delays. The political question is whether Tan's China file is read as a disqualifier, a manageable issue, or a bargaining chip in a renegotiation that may already be underway. The economic question is whether Intel can afford to lose $8 billion of subsidy in a year when its foundry business is still burning cash.

The foundry split: what it actually solves

The case for splitting Intel into a design company and a foundry company is the same case that has been made for five years and has gained weight with each delayed process node. Foundry customers — Apple, AMD, Nvidia's competitors in networking and automotive, a long tail of fabless designers — want an independent, neutral, technically credible second source to TSMC. Intel's foundry, entangled in Intel Products, has not been able to offer that. A clean separation would unlock external wafer demand, justify a different cost-of-capital conversation for the fabs, and let the design business compete with AMD without carrying the burden of capital expenditure that is two generations behind Samsung and four behind TSMC.

The case against is the case Intel's management has been making: that the synergies between leading-edge design and leading-edge process are the entire point of an integrated manufacturer, that splitting them destroys the moat, and that the foundry business, as a standalone, will not generate the returns to justify its capex without anchor customers Intel Products is currently supplying. Tan's first staff note tries to thread the needle. "In areas where we are behind the competition we need to take calculated risks to disrupt and leapfrog," he wrote. "In areas where our progress has been slower than expected we need to find new ways to pick up the pace." Read in one direction, that is a break-up memo. Read in another, it is a capital-discipline memo. The board has presumably asked him which, and presumably has not been given a straight answer, because a straight answer at this stage is a negotiating position he cannot afford to give away.

The capital backdrop: policy fog and a VC market that has stopped pretending

Tan is not negotiating in a vacuum. Logan Bartlett, presenting a Redpoint LP update for March 2025, described Morgan Stanley post-conference sentiment as the most negative in a decade outside the COVID onset — driven, in Bartlett's telling, by policy uncertainty. The US Policy Uncertainty Index recently hit its second-highest level since 1985, exceeded only by early COVID. Consumer sentiment is sliding and delinquency probability is rising. That is the macro in which Intel has to make its case to capital markets, to the White House, and to a customer base that has spent five years quietly hedging.

The venture market around Intel is sending its own signals. Bartlett framed the cycle as one in which "smart endowments and pensions have walked from bloated $3-6B funds," but "sovereign wealth and Middle East capital has replaced them as a 'rounding error,' keeping mega-funds alive." The structure is the same one Sam Lessin named in another segment: a market in which the average discipline breaks because "the average of infinity and zero is infinity." Figure's $40 billion pricing in a recent round, Bartlett argued, has no explanation other than special-purpose vehicles in which managers are writing checks beyond their fund capacity. That is the same pattern, in microcosm, as the SPV era — late-cycle capital formation dressed up as strategy.

The sustaining-versus-disruptive read

The most useful frame for the Intel decision comes from a debate on the same broadcast between Seth Rosenberg of Greylock and Sam Lessin, the latter making the more provocative case. Lessin's read is that the middle of the software stack is being hollowed out: hyperscalers are bundling AI APIs for free, and the sub-$10 billion public software company is no longer a desirable destination for capital. "No one wants to be a $10B public company anymore," Lessin said. "You either want to be a hyperscaler or a one-person billion-dollar company." Applied to Intel, that is a more radical statement than it first appears. Intel, at a market cap of roughly $130 billion on a good day, is exactly the kind of middle that the new capital cycle is penalising. The strategic question Tan faces is not whether Intel can be a $300 billion company again — that math does not work in the current cycle — but whether it can attach itself to a hyperscaler-scale ecosystem or, alternatively, find the narrow set of customers and end markets where an integrated mid-cap manufacturer can still earn a return on capital.

Bartlett's adjacent point about Apple is instructive even though Apple is not Intel's problem. Apple, with a roughly $3.2 trillion market capitalisation and more than $1 trillion returned to shareholders over the last fourteen years, is a "deterministic culture, not a probabilistic culture," and is therefore structurally misaligned with AI's iteration-and-proliferation requirements. The same critique applies, more sharply, to Intel. The question is whether Tan can rebuild a culture that is comfortable shipping product that is not perfect, and whether the board will let him.

The stake for Washington, for the foundry, and for Tan personally

The stakes in the next twelve months are unusually clean. For the Trump administration, the question is whether to treat the CHIPS Act as a sunk cost and a political embarrassment, or as a continuing industrial-policy tool that can be renegotiated into something faster. For Intel's foundry customers, the question is whether Tan's separation thesis produces a real second source within a planning horizon they can use, or whether they should accelerate their TSMC concentration now. For Tan personally, the question is whether the 3,000% track record at Cadence was a function of his own decisions or of a cycle that has since closed, and whether his history as a cross-border dealmaker becomes an asset or a liability in the political environment he is now operating in.

The honest answer at this stage is that the answer does not yet exist. What exists is a board that has made a credible hire, a market that has given that hire one quarter of grace, a government that has signalled it is willing to use CHIPS funding as leverage, and a foundry business that has not yet demonstrated it can be run as a standalone. The shape of the answer will be visible by the middle of 2027. Until then, the year Lip-Bu Tan has been given is also the year Intel is being given — and the rest of the industry will be measuring it.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://www.youtube.com/watch?v=eZdnwjHpIIQ