The Apollo Exit That Could Redirect Billions Into Western Chip Manufacturing — and Washington Is Paying the Bill

The honest answer so far: almost nobody external. Intel Foundry generated $5.4 billion in intersegment revenue in Q1 — but that figure is almost entirely Intel making chips for itself. The actual external foundry revenue was a fraction of that total. The Data Center and AI segment pulled in $5.1 billion, and the PC division brought in $7.7 billion, but those are Intel's own product lines, not outside customers choosing Intel Foundry over TSMC.

That absence of named external customers in this earnings release is itself a signal. Intel has been promising for years that it could compete as a contract chipmaker. The Fab 34 buyback — $14.2 billion to take full ownership of the Ireland factory, originally structured as a joint venture with Apollo in June 2024 — shows Intel is doubling down on the bet. But doubling down and winning customers are two different things.

The real tell for the next few quarters isn't revenue growth in Intel's own product lines. It's whether any major external logo signs a foundry commitment. Until that happens, the catering business is still serving only the parent restaurant.

The biggest one-time charges were $4.1 billion in restructuring costs and a $3.9 billion goodwill (the premium paid for a company above its book value) impairment on Mobileye. Those are painful but backward-looking. The forward-looking risk is structural: Intel carries $43 billion in long-term debt, holds $17.2 billion in cash and $15.5 billion in short-term investments, and just took on a $6.5 billion bridge loan to fund the Fab 34 buyback.

Bridge loans, by definition, need to be refinanced. That refinancing window will open while CHIPS Act disbursements are still being parceled out on a milestone-gated schedule controlled by the Commerce Department. Intel's capex plan is partially hostage to bureaucratic sign-off timelines that could slip independently of Intel's own execution.

Say the subsidies arrive late and the bridge loan comes due while interest rates are still high — or under a less chip-friendly administration. Intel would face a refinancing crunch that the current stock price doesn't account for. The margin for error in this specific 12-to-18-month corridor is razor-thin.

Intel recognized $629 million in capital-related incentives and another $176 million in other grants in Q1, per its filings. That's over $800 million in a single quarter from government programs designed to bring chip manufacturing back to the U.S. And its allies. Intel also spent $3.6 billion on capital expenditures in the quarter.

The catch: these subsidies aren't a blank check. CHIPS Act disbursements are milestone-gated by the Commerce Department, meaning Intel must hit specific construction and production benchmarks before each tranche is released. If a factory buildout hits a permitting delay or an equipment shortage, the subsidy timeline slips — even if Intel's own execution is on track.

That bureaucratic dependency creates a risk most sell-side models don't capture. Subsidy programs can also hit political pushback, require matching private investment, or get reshaped when a new administration takes over. Intel stands to gain the most from bipartisan support for U.S. Chip-making. But political agreement can fall apart faster than a factory goes up.

Intel took a $3.9 billion goodwill impairment on Mobileye. That's Intel publicly admitting its self-driving car unit has lost significant value since the original deal. Mobileye was supposed to be Intel's ticket to the autonomous vehicle revolution. Instead, it's become a drag on the balance sheet.

The writedown suggests the commercial timeline for self-driving technology is stretching further into the future than Intel originally projected. And the ripple extends beyond Intel. Mobileye itself is publicly traded (MBLY), and a parent company marking down its stake puts direct pressure on the subsidiary's market valuation. Aurora Innovation, another public self-driving company, faces the same tough questions: how long its cash will last and when it can actually sell a product.

Intel has deep pockets and a long time horizon. If even Intel is admitting the bet isn't paying off on schedule, smaller players with less cash face even harder questions about whether their own stock prices can hold up.

Apollo invested in Intel's Fab 34 joint venture in June 2024. Less than two years later, Intel bought back Apollo's 49% stake for $14.2 billion, completing the deal on April 1, 2026. Here's how the deal worked: a PE firm helped fund a factory, and the chipmaker bought them out once the plant proved its value. By chip-industry standards, that exit looks fast. The exact financial return hasn't been made public.

The structure itself is the story. Building a cutting-edge fab costs tens of billions and takes years. Governments can subsidize part of the bill, but they can't fund all of it. If private equity sees the Apollo-Intel deal as a playbook it can repeat, a lot more money could flow into Western chip factories from a source that barely showed up before.

Amazon is pouring $25 billion into its partnership with Anthropic, and AI infrastructure spending is accelerating across the industry. Data centers need more than just GPUs — they need CPUs, networking chips, and the factories to make them all. A proven PE co-investment model for fabs could meaningfully expand the capital pool available for Western semiconductor buildout.