For starters, we realized we did not have a clear picture on just how much money Intel needs. Our thesis is that Intel needs to survive long enough to get its 14A manufacturing process into volume production. If that works as promised (if), and if they can reach economical yields (a bigger if), then the company can be seriously competitive again and maybe even have a shot at building its foundry business. As far as we can tell, the current roadmap calls for 14A to go into first production in 2026. Let’s assume that in production in 2026 means that the full benefits of 14A do not start to show in numbers until 2027. So the company has to keep the lights on for two and a half years.
A big part of the reason that last month’s earnings miss sparked a panic was the company needs to generate sufficient product revenue to fund their yawning capex needs. Falling revenue estimates call into question the viability of the entire enterprise.
The Street currently expects 2024 revenue of $52 billion and 2025 revenue of $56 billion. Both of those numbers look too high to us. So for simplicity’s sake, let’s say they do $50 billion in revenue this year and next, and let’s also assume gross margins come in at 40% this year and next. Let’s assume that things pick up in 2026 and revenue gets to $55 billion on 45% gross margins. Adding that all up comes to about $65 billion in gross profit.
The company guided to $20 billion in opex this year and $17.5 next. Let’s assume they maintain that cost discipline into 2026 as well. That adds up to $55 billion in total opex over three years. On top of that they guided to $26 billion in capex this year, $22 billion in 2025 and let’s assume the same amount in 2026. This work out to $71 billion in opex. So over the next three years the company will bring in $65 billion, but needs $126 billion, a funding gap of $61 billion.
So that’s a big number, but we need to make some adjustments. The government is expected to chip in a fair amount via the CHIPS Act, and a fair portion of these expenses can be paid in stock. The company provided both gross and net capex figures which deduct government funds and stock comp. Net capex is $12 billion in 2024 and $13 billion in 2025. This bring down the total funding gap to $28 billion.
That is a manageable number, but still large. Where can Intel find that much money? Most people we speak with seem to think that the US government is going to foot the bill. We think that is very unlikely in an election year, and while we cannot even guess as to what the political scene looks like next year, a full bailout of Intel seems unlikely to come from either party. Moreover, we actually think a direct government subsidy is a really bad idea. For starters, the US is bad at this and risks creating a stunted company. For example, IBM gets a fair amount of government support and to be diplomatic has not demonstrated the commercial prospects that many may have hoped from such support. On top of that, the goal of such a subsidy would be to “bring leading edge semiconductor manufacturing back to the US”, but that policy aim totally misses the point. The US already has the ability to manufacture leading edge semis in the US, what is lacking is the ability to do that manufacturing economically. Intel can produce chips with EUV today, they can produce 14A chips today, but they cannot do it profitably at volume. We can debate the degree to which the US government wants to subsidize some base level of manufacturing capacity (and Ben Thompson’s ideas about subsidizing demand for leading edge semis sounds like a good idea), but at this stage support for Intel is largely about supporting their viability as a commercial entity not as a national security asset.
This leads back to our suggestion that the best path for Intel is to seek support from potential foundry customers. The group could include some combination of Apple, Amazon, Meta, Google, or Broadcom. Add to that list some Intel competitors like AMD, Nvidia or Qualcomm. Yes, this would be messy. Yes, there are all kinds of governance complications. But it is a viable option. For this group, $28 billion is not an impossible amount. Of course, none of them are likely to do this on their own, and maybe the government needs to step in and encourage a bit of “patriotism”, but it is also in all these companies’ interest to have a viable alternative to TSMC some day. Ultimately, Intel’s needs are large, but not so large as to be unbridgeable. There is a workable solution out there.
Leave a Reply