Intel 10% by 2030 still?

[OneEng]

All that has changed based on the lack of traction. the goal was 15B in external revenue. Add that to internal and IFS makes sense.

and If intel is going to claim packaging as foundry, then you need to add in ASE and AMKOR.

Intel needs 15B in external revenue to make the finances make sense. The plan was 2030 for this. There is not line of sight to get there yet. Our model is that if ALL of the speculated commitments came true, they are about halfway to goal.

The intermediate goal should be to be a top 10 foundry in external revenue (wafer fab) and Top 5 packaging house in external revenue.

I guess what I wonder is how much did Intel intend to spend in order to reach the 15B in external revenue? I am guessing that Intel can not service 15B in additional external revenue without increasing the number of lines and fabs.... is that correct?

So to answer your question: Intel (and AMD) needs to prioritize revenue, profit, and the quality of revenue above everything else. Market share is still important, but it is becoming increasingly misleading. If Intel can generate $30 billion in annual revenue while earning $10 billion in net profit, isn’t that far better than Intel’s 2025 result of $52.9 billion in revenue with only $300 million in net profit?

Agree completely.

Just FYI, Adjusted for one time charges, 2024 Operating Loss would be also mid 50s% only.

Now lets ask what happened in 2023,2024 & 2025? Intel started outsourcing a significant % of its wafers to TSMC. Intel management said about 30% of outsourced wafer at a peak. Also they had high node ramping\starting costs for Intel 4, Intel 3, Intel 18A due to accelerated node development process.

That ramp has slow down and its effect on operating loss will go away as they will use incremental nodes for future products 18A ->18A-P->18A-U etc, eventually 14A ramp cost will show up but not before 2027 per CEO commentary.

Now with Panther Lake wafers coming back mostly internal and Nova lake will be improving on that, so this operating loss % will come back down over time. As yield improves on 18A & Intel 3 and as volume ramps up on 18A & Intel 3, I expect this loss to move towards to 0% (break even). This is evident in Q1'2026, margin improved 10% QoQ. We will see how this metric progresses over the next 8 quarters & if Intel can achieve 0% as originally targeted. (Exiting 2027 with Intel Foundry breakeven on operating profit basis)
View attachment 4598

In 2021/2022, Intel foundry had 0% gross margin and still negative operating margins because it was being operated as an IDM foundry (Wafers sold to products at cost with no profit motive). When Products team started souring wafer from TSMC, Intel Foundry gross margin turned negative (mainly because of underutilization & high startup costs), ofc operating margin became even more negative. Now with foundry charging market prices to Intel Products team, Foundry P&L is mainly dependent on wafer volume. It does not matter internal or external. If they can completely decouple from TSMC and still effectively compete with 100% wafer at Intel Foundry, that is a good thing for Intel Foundry P&L.

With TSMC capacity constrained due to being too conservative, I think more and more customers are looking at Intel Foundry seriously and that threshold of wafer volume required for Intel Foundry to flourish is not too far away.

While I would agree that Intel's "Have it All ways at once" approach was God Awful Expensive, and that moving production back to Intel from TSMC would be a MORE profitable (or less loss) path, it isn't like the spending at IFS ends there. There will be insane amounts of money to be spent on new equipment to keep up with TSMC (I know they have already purchased some High NA equipment to try to regain the lead in the future) and also insane amounts of money spent on new processes.

Your point about the insatiable demand for high end chips is spot on though. This puts even more emphasis on Intel not trying to eat the whole pie at once. They need to create a profitable business with sustained growth. The rest will take time, but this should be the benchmark they are first targeting.

 

[hist78]

I guess what I wonder is how much did Intel intend to spend in order to reach the 15B in external revenue? I am guessing that Intel can not service 15B in additional external revenue without increasing the number of lines and fabs.... is that correct?

Last year, Intel CEO Li‑Bu Tan stated that Intel would need additional capital spending on 18A if external customers arrive, although Intel’s internal demand alone is sufficient to bring 18A into high‑volume manufacturing.

He also said that 14A will be much more expensive and will require both internal demand and meaningful external foundry customers in order to bring 14A to life.

Simply put, Intel needs a lot more cash to support the high level of capital spending required to remain competitive as an IDM, even after already spending a massive $125 billion in CapEx from 2020 to 2025.

Intel’s 2026 CapEx will be around $14–15 billion, while TSMC’s 2026 CapEx will be between $52 billion and $56 billion.

Although bringing more outsourced production back into Intel Foundry may improve production volume, efficiency, and cost, Intel still faces a serious question: Where will the money come from to sustain the IDM 2.0 model?

With the current fast pace and large scale of AI deployment, Intel’s customers cannot afford to wait for Intel to figure this out. And Intel’s competitors certainly will not wait for Intel either.

Li‑Bu is likely debating whether Intel Product can rely solely on Intel Foundry to stay competitive, especially when Intel Product is the division generating the majority of Intel’s revenue and profit. If the answer is no, or even uncertain, then Intel must continue using TSMC in order to expand and manage risks (for both technical risks and financial risks).

 

[Xebec]

ARM/RISC CPUs are already chipping away at x86 growth and, by my own estimate, have surpassed x86 in terms of value and revenue. In many cases, Intel and AMD are facing competition from companies with in-house semiconductor divisions that have advantages in price, cost, features, performance, and strategic importance.

It’s hard for Intel to maintain ASPs and margins when those companies prioritize their internal semiconductor divisions for high-impact, high-value needs and leave Intel with the lower-margin, less exciting business.

Agreed - I think they've surpassed x86 too.

I was just thinking that Intel still "has" the laptop market -- but they can do a lot more to slow down or even halt the continued exodus here to other devices. Until the recent "AI caused parts shortage", Intel even had an opportunity to expand a bit with good mobile gaming solutions like Panther Lake. (They could try to take some AMD and ARM owned console gaming market share with the right advertising and products..)

I think Intel also needs to one-up AMD here by getting some of those ARM and RISC-V products into it's fabs -- so they can at least get a portion of the revenue that AMD can't get on those products. AMD IMO is still the worst threat to Intel as AMD will even go for 'phyrric' victories just to beat Intel. (see the IP transfer of Zen 1 to the Chinese - reducing the x86 TAM in China permanently, or offering 10% of the company to OpenAI to get their business - locking competition out). This makes AMD very dangerous to Intel's long term health.

 

[MKWVentures]

I guess what I wonder is how much did Intel intend to spend in order to reach the 15B in external revenue? I am guessing that Intel can not service 15B in additional external revenue without increasing the number of lines and fabs.... is that correct?

I think the plan was pretty clear. 2 fabs in Arizona, 2 Fabs In Ohio.... then israel then Germany. They committed to have 4 new fabs by 2026 minimum. They have one fab now, 52 doesnt hit full planned capacity until end of year. As LBT clearly discussed, that plan put Intel in a very dangerous position (hence the stock price was below book value until LBT fixed it).

high level model was 2-3 fabs on leading edge node. That is still the only model for financial success.... if you do not care about finances (Im talking to you Elon), you can do whatever you want LOL.

 

[dkr1986]

While I would agree that Intel's "Have it All ways at once" approach was God Awful Expensive, and that moving production back to Intel from TSMC would be a MORE profitable (or less loss) path, it isn't like the spending at IFS ends there. There will be insane amounts of money to be spent on new equipment to keep up with TSMC (I know they have already purchased some High NA equipment to try to regain the lead in the future) and also insane amounts of money spent on new processes.

Although bringing more outsourced production back into Intel Foundry may improve production volume, efficiency, and cost, Intel still faces a serious question: Where will the money come from to sustain the IDM 2.0 model?

Intel generates roughly $10B in Cash flow from Operations in 2025 even with all their problems. The drag on this metric due to Foundry not being profitable is roughly >$8B. So if Intel Foundry breaks even on EBIT basis, Intel could generate around $19B in cash flow from operation. If they fix Foundry economics (with increased utilization [Intel 7 & 3 sold out for AI CPU demand, Intel/UMC 12 starts making money in 2027 putting those older fabs to work, Intel 18A ramped in fab 52 & sold out or fully allocated to internal & some external demand], rolling over from the high node startup cost, maybe even external customers with prepays) while taking part in the AI DC CPU TAM for Intel Products increasing the total revenue, they can comfortably sustain a capex of ~$20B each year starting in 2028 (Intel's capex was ~$25B during accelerated node dev + capacity expansion, so a ~$20B spend during normal cadence is likely more than enough). They have enough cash buffer and partner contributions/offsets to stay in the game until end of 2027. Also they speed ran some of the spends like building up the shells under PG (They have shell space in Israel Fab 38, Fab 62, Ohio Fabs are still being constructed) & bought up equipment.

At $25K per price and 20k wafer per month (roughly one Intel fab), the revenue is 25kx20kx12 = $6B. Since Intel's products are split across multiple nodes, Fab 28 & Fab 42 in Israel & Arizona (Intel 7), Fab 34 in Ireland (Intel 4/3) & Fab 52 (Intel 18A) in dedicated to Intel. Considering slow down in consumer products, Fab 62 is likely not needed for Intel for another year or so. Or they can wean customers off Intel 7 and fully upgrade Fab 42 to Intel 18A to increased 18A capacity. So Intel can ramp external client in Fab 62. Fab 38 is also fully built and shelled, they can ramp that one too. These fabs only need equipment spend which Intel has said most of their capex is going to now.

TSMC is building shells, upgrading exiting fabs and expanding capacity into already built fabs to service multiple customers, Intel Foundry is only servicing Intel Foundry team and some incremental external customers, so I don't think Intel needs to spend as much as TSMC (only on node development spends are directly comparable). Also staying competitive on node process means spend in R&D (Operating expense) and to buy equipment (Capital expense).

I did some modelling of these things with some rough assumptions Lot of simplifications here to simplify the forecasts but roughly captures the trend. Some assumptions are modified further as shown in the second picture. My assumption is $10B in external foundry revenue in 2030 with $6B coming from wafer orders and $4B coming in from Advanced packaging solutions. I am only using ChatGPT to save working with an Excel worksheet, All the assumptions are mine, AI is just calculating the numbers (in lieu of Excel)

As I said above, I redid my calculations with more reasonable assumptions. I also did Free Cash Flow estimate this time accounting for capex that need to be spent.

My post on X if you want to see how it progressed.

https://twitter.com/x/status/2057105709850861712

 

[dkr1986]

Before you all question this assumption for DCAI, Let me explain. I assumed $25B of DCAI revenue based on 10% QoQ growth over the Q1'26 DCAI revenue of $5.3B (they grew 8+% QoQ from Q4'25). Intel management said, Q1'26 is the most capacity constrained quarter and more capacity coming online on each following quarter. This is based on huge demand for DC CPUs in AI market. So 5.3 x 1.1^3 = ~$25B DCAI revenue for 2026. Intel is the only company right now with excess fab capacity (Intel 7 due to consumer products moving to Intel 18A & more tool installation in Fab 34 for Intel 3) to capture this incremental demand in my opinion. The DCAI TAM is coming from AMD's investor relation presentation adjusted for revised DC CPU TAM of $120B in 2030. So I am not assuming 70% CPU share explicitly, it turned out that way. Either my 10% QoQ is too optimistic or the TAM is underestimated in 2030.