I tend to think on future out-of-the-world bluesky scenarios:
1) I'm curious on your thoughts for their goals in Silicon Photonics. If they really want to compete, it seems moving to 300mm and 22nm or below is necessary. Will they really pull the trigger?
2) Also curious on neuromorphic front, if it would be smart for them to acquire some memory IPs (like weebit or 4DS) to round out their analog portfolio?
3) If GaN-on-Si appears to be big down the road, how can they compete without 300mm?
4) What's the outlook on microfluidics? Are there any interesting applications in the field of biotech that might make this piece of business important?
The MOU with SMART Photonics directly addresses this. Instead of building out 300mm/22nm capabilities internally (which would be a massive capital investment, in the billions, and deviation from their strategy), X-FAB is pursuing a partnership approach. This makes strategic sense - they get to participate in silicon photonics without abandoning their successful specialty manufacturing model.
X-FAB's attendance at the Optical Fiber Communications Conference confirms they're serious about photonics, but their approach seems to be leveraging partnerships and their existing expertise rather than "pulling the trigger" on ultra-advanced nodes. This preserves their capital efficiency while still giving them market entry.
I believe this partnership approach is smarter than going all-in on advanced nodes which would put them in direct competition with TSMC, Samsung and Intel - a battle that would be extremely difficult to win given their relative size.
2.
The Attopsemi partnership for I-fuse OTP on their XH018 platform gives us insight into X-FAB's memory strategy. Rather than acquiring companies like Weebit or 4DS outright, they're following their proven playbook of strategic partnerships. (they love being a partner)
This approach gives their customers memory options without requiring fundamental changes to X-FAB's manufacturing processes (their announcement specifically mentions "no mask adders to the X-FAB core process").
For neuromorphic applications specifically, their strong analog/mixed-signal expertise is already well-positioned without needing radical changes. If they continue this partnership strategy for neuromorphic-specific memory solutions, they could expand their capabilities while maintaining their focus on what they do best - specialty manufacturing with proven processes.
3.
I think their approach to SiC provides clues. The XSICM03 announcement shows they're achieving significant improvements (25% smaller cell pitch, 30% more die per wafer) on their existing wafer sizes.
GaN-on-Si is currently still predominantly manufactured on 150mm and 200mm wafers industry-wide - exactly where X-FAB specialises. Their strategy appears to be optimising these existing wafer sizes rather than rushing to 300mm.
Their partnership with Soitec for SmartSiC suggests they may pursue similar material partnerships for GaN rather than developing everything internally. This would let them leverage external expertise while contributing their manufacturing knowledge.
Given their "XbloX" platform approach that accelerates development by up to nine months, they're showing they can compete through process efficiency rather than wafer size alone.
4.
The CEO's quote when receiving the Onderneming van het Jaar award (fancy name!) mentioned "intelligent medical applications that address the needs of a growing and ageing world population," suggesting they see medical applications as a strategic focus(?)
Their partnership with Attopsemi specifically mentions new opportunities in the medical sector, showing this vertical is on their radar.
X-FAB already has experience with MEMS (micro-electro-mechanical systems), which uses similar manufacturing techniques to microfluidics. This existing expertise positions them well for biotech applications like lab-on-chip diagnostics, which wouldn't necessarily require cutting-edge nodes.
The biotech application space aligns perfectly with X-FAB's strength in specialty manufacturing for critical systems where reliability trumps raw performance - exactly the kind of niche where their current business model thrives.
---
Again, that's all I can see with some research I did today, but I cannot claim to be a semis expert, so I'm more than happy to learn more from you.
I'm focused on the new write-up for Wed at the moment. I'll try to find time next week, to see what I can answer to your questions, since I need time to refresh my memory on Xfab and also understand in detail the scenarios you mention.
A beautifully written piece. But to me, from what I just read, XFAB looks like a no-growth company, a essentially a bond substitute, with risks higher than those of a bond.
"No-growth" feels like a strange label - at 6.28x earnings with a solid moat, that comparison isn't crazy. But the numbers suggest something different: 22.6% revenue growth, 200mm lines at full capacity, and customers prepaying €275.9M just to secure future production.
I would think of them less as a bond and more like the owner of a very specific bridge that everyone needs to cross - it might be old, but until someone builds a better one (and gets it certified for automotive use), those toll payments keep flowing.
The "book-to-bill ratio" – simply the ratio of new orders to actual sales – tells a fascinating story. For their traditional business, they're getting $1.05 in new orders for every $1 of current sales. But in their new silicon carbide chip business (crucial for electric vehicles), they're only getting 38 cents in new orders for every dollar of sales."
So is the revenue growth, then, due to a forward pull from pre-orders?
In short, the evidence shows X-FAB's revenue trends reflect real operational performance rather than financial engineering through prepayments.
Here's why:
Proven by evidence:
- Prepayments ($316M total) are held as liabilities, not revenue (Interim Report H1 2024, p.25, Note 2.5.6.15)
- Revenue is recognized based on production over time or delivery (Interim Report H1 2024, p.17, Note 2.5.6.1)
- Year-over-year changes reflect actual business shifts (Interim Report H1 2024, p.4):
+10% automotive
-11% industrial
-18% medical
We can reasonably infer:
- The mixed performance across segments suggests authentic reporting rather than managed growth (Interim Report H1 2024, p.4 segment breakdown)
- The regional divergence (China +55%, EU/US weakening) aligns with broader market conditions (Q3 2024 earnings call transcript, p.4)
- As we said in the article, there is a transition, with traditional segments declining while newer areas (180nm) show strength (Q3 2024 earnings call transcript, p.4)
The total prepayments break down into:
- Direct prepayments for future wafers: $40.6M
- Capacity reservation deposits: $275.9M
(Both figures from Interim Report H1 2024, p.25, Note 2.5.6.15)
So while X-FAB has substantial prepayments from customers, these don't create artificial revenue growth - they're more an indicator of customer commitment to future capacity than a current period revenue driver.
Think risks have been totally overblown on this name. China revenue was single figure in 2023, so mid teens in 2024? They will win the EV war so I don't see how they can compete without it. I don't see trade restrictions being focused on these lower value components. Tariff risk is real but in their Q3 letter Night Watch IM pointed out weakening FX helps to increase competitiveness of the German plants. Germany also reached 64% renewables in 2024, could get to 75% by 2026. Power prices only going one way - down.
Biggest threat looks like SiC and getting eaten by bigger fish, but that seems to be normalising.
Just the fact XFAB is receiving so many prepayments (not only from Melexis) shows the risk/reward on the expansion isn't quite what it seems on the surface.
Thanks for this - really helpful perspective that made me think!
You're right, I probably overstated some of these risks. That China number is a good example - the 55% growth sounds huge, but when you're starting from single digits, even that kind of growth keeps the total exposure pretty manageable.
Good catch on the German energy situation too. I hadn't realized renewables had hit 64% and where power prices were heading. Combined with the currency advantages you mentioned from Night Watch's analysis, the German manufacturing cost picture looks better than I painted it.
On prepayments - it is a telling signal. When customers put down €275.9M in deposits, they're essentially helping fund X-FAB's expansion. That does change the risk profile of their capex program.
On SiC - yeah, fair point about normalization. I might have focused too much on the current dip versus the longer-term story.
Really appreciate you taking the time to add this context. These kinds of discussions help build a better picture.
Nice writeup! For what its worth I was referring to the market's perception of risks rather than your piece. What's interesting is it seems to have derated down to essentially a cyclical Auto OEM/tier 2 supplier valuation at >8 PE, >5 EBIT as if trading into a downturn and profits will tumble 30-40%, while comparables are 50-200% higher. With the rest of the industry cutting capex, you could say the market is treating this expansion as a complete money furnace when they are still guiding for higher profits next year. So this resolves itself one way or the other anyway: expansion doesn't happen - minimal BS impact and shareholder returns, multiples improve. Expansion happens - shares rerate.
Thanks - this is really helpful framing on the valuation.
I take it you are quite bullish on X-FAB.
My understanding is that essentially the stock is priced for disaster but has two potential positive outcomes:
- Either they successfully expand (good)
- Or they don't expand and return capital (also good)
Correct?
Two questions if you don't mind:
1. Which companies do you look at as the most relevant comps? The peer group is tricky since X-FAB sits between pure foundries and auto suppliers.
2. As fleets transition from ICE to EV, does this generally mean more demand for X-FAB's type of analog/mixed-signal chips? Trying to understand if electrification is a tailwind for their core competencies.
Yes I mean if they genuinely see little RoI they will surely cancel what remaining capex they can. But I think potential problems are much more post expansion (scaling up/economics etc), because this is a structurally constrained market, as Covid showed.
1. From what I can see there are two groups of comps: the pure play mixed-signal/analog foundries (Vanguard Int Semi, Tower Semi, Hua Hong Semi, DB HiTek), and the larger integrated designers/manufacturers that have a focus on autos (the Infineons, SMTMicro, Renesas).
Interesting they're most optimistic on e-braking, if that's already a large business at Melexis (don't know, haven't checked) that is very good news for XFAB.
This is being slammed from tariffs and rising euro. US exposure for XFAB/Melexis is 5-10%, another few % indirectly through EU customers. What are people's thoughts? I think orders can stay strong on China growth + UK being spared by Trump.
I tend to think on future out-of-the-world bluesky scenarios:
1) I'm curious on your thoughts for their goals in Silicon Photonics. If they really want to compete, it seems moving to 300mm and 22nm or below is necessary. Will they really pull the trigger?
2) Also curious on neuromorphic front, if it would be smart for them to acquire some memory IPs (like weebit or 4DS) to round out their analog portfolio?
3) If GaN-on-Si appears to be big down the road, how can they compete without 300mm?
4) What's the outlook on microfluidics? Are there any interesting applications in the field of biotech that might make this piece of business important?
You seem to be more knowledgeable than me in semiconductors.
I'll try to answer with whatever little I know.
( most of the information I refer to is within https://www.xfab.com/news )
1.
The MOU with SMART Photonics directly addresses this. Instead of building out 300mm/22nm capabilities internally (which would be a massive capital investment, in the billions, and deviation from their strategy), X-FAB is pursuing a partnership approach. This makes strategic sense - they get to participate in silicon photonics without abandoning their successful specialty manufacturing model.
X-FAB's attendance at the Optical Fiber Communications Conference confirms they're serious about photonics, but their approach seems to be leveraging partnerships and their existing expertise rather than "pulling the trigger" on ultra-advanced nodes. This preserves their capital efficiency while still giving them market entry.
I believe this partnership approach is smarter than going all-in on advanced nodes which would put them in direct competition with TSMC, Samsung and Intel - a battle that would be extremely difficult to win given their relative size.
2.
The Attopsemi partnership for I-fuse OTP on their XH018 platform gives us insight into X-FAB's memory strategy. Rather than acquiring companies like Weebit or 4DS outright, they're following their proven playbook of strategic partnerships. (they love being a partner)
This approach gives their customers memory options without requiring fundamental changes to X-FAB's manufacturing processes (their announcement specifically mentions "no mask adders to the X-FAB core process").
For neuromorphic applications specifically, their strong analog/mixed-signal expertise is already well-positioned without needing radical changes. If they continue this partnership strategy for neuromorphic-specific memory solutions, they could expand their capabilities while maintaining their focus on what they do best - specialty manufacturing with proven processes.
3.
I think their approach to SiC provides clues. The XSICM03 announcement shows they're achieving significant improvements (25% smaller cell pitch, 30% more die per wafer) on their existing wafer sizes.
GaN-on-Si is currently still predominantly manufactured on 150mm and 200mm wafers industry-wide - exactly where X-FAB specialises. Their strategy appears to be optimising these existing wafer sizes rather than rushing to 300mm.
Their partnership with Soitec for SmartSiC suggests they may pursue similar material partnerships for GaN rather than developing everything internally. This would let them leverage external expertise while contributing their manufacturing knowledge.
Given their "XbloX" platform approach that accelerates development by up to nine months, they're showing they can compete through process efficiency rather than wafer size alone.
4.
The CEO's quote when receiving the Onderneming van het Jaar award (fancy name!) mentioned "intelligent medical applications that address the needs of a growing and ageing world population," suggesting they see medical applications as a strategic focus(?)
Their partnership with Attopsemi specifically mentions new opportunities in the medical sector, showing this vertical is on their radar.
X-FAB already has experience with MEMS (micro-electro-mechanical systems), which uses similar manufacturing techniques to microfluidics. This existing expertise positions them well for biotech applications like lab-on-chip diagnostics, which wouldn't necessarily require cutting-edge nodes.
The biotech application space aligns perfectly with X-FAB's strength in specialty manufacturing for critical systems where reliability trumps raw performance - exactly the kind of niche where their current business model thrives.
---
Again, that's all I can see with some research I did today, but I cannot claim to be a semis expert, so I'm more than happy to learn more from you.
Very interesting questions, thank you.
I'm focused on the new write-up for Wed at the moment. I'll try to find time next week, to see what I can answer to your questions, since I need time to refresh my memory on Xfab and also understand in detail the scenarios you mention.
A beautifully written piece. But to me, from what I just read, XFAB looks like a no-growth company, a essentially a bond substitute, with risks higher than those of a bond.
"No-growth" feels like a strange label - at 6.28x earnings with a solid moat, that comparison isn't crazy. But the numbers suggest something different: 22.6% revenue growth, 200mm lines at full capacity, and customers prepaying €275.9M just to secure future production.
I would think of them less as a bond and more like the owner of a very specific bridge that everyone needs to cross - it might be old, but until someone builds a better one (and gets it certified for automotive use), those toll payments keep flowing.
I guess I was zeroing in on this: "
The "book-to-bill ratio" – simply the ratio of new orders to actual sales – tells a fascinating story. For their traditional business, they're getting $1.05 in new orders for every $1 of current sales. But in their new silicon carbide chip business (crucial for electric vehicles), they're only getting 38 cents in new orders for every dollar of sales."
So is the revenue growth, then, due to a forward pull from pre-orders?
Very good question.
Had to dig a bit deeper on that one.
In short, the evidence shows X-FAB's revenue trends reflect real operational performance rather than financial engineering through prepayments.
Here's why:
Proven by evidence:
- Prepayments ($316M total) are held as liabilities, not revenue (Interim Report H1 2024, p.25, Note 2.5.6.15)
- Revenue is recognized based on production over time or delivery (Interim Report H1 2024, p.17, Note 2.5.6.1)
- Year-over-year changes reflect actual business shifts (Interim Report H1 2024, p.4):
+10% automotive
-11% industrial
-18% medical
We can reasonably infer:
- The mixed performance across segments suggests authentic reporting rather than managed growth (Interim Report H1 2024, p.4 segment breakdown)
- The regional divergence (China +55%, EU/US weakening) aligns with broader market conditions (Q3 2024 earnings call transcript, p.4)
- As we said in the article, there is a transition, with traditional segments declining while newer areas (180nm) show strength (Q3 2024 earnings call transcript, p.4)
The total prepayments break down into:
- Direct prepayments for future wafers: $40.6M
- Capacity reservation deposits: $275.9M
(Both figures from Interim Report H1 2024, p.25, Note 2.5.6.15)
So while X-FAB has substantial prepayments from customers, these don't create artificial revenue growth - they're more an indicator of customer commitment to future capacity than a current period revenue driver.
Thank you for this great question!
( H1 report referenced : https://www.xfab.com/securedl/sdl-eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpYXQiOjE3Mzc0NDE4OTMsImV4cCI6MTczNzUzMTg5MywidXNlciI6MCwiZ3JvdXBzIjpbMCwtMV0sImZpbGUiOiJmaWxlYWRtaW4vWC1GQUIvSW52ZXN0b3JfUmVsYXRpb25zL1JlcG9ydHMvWC1GQUJfSGFsZi1ZZWFyX1JlcG9ydF8yMDI0X0VORy5wZGYiLCJwYWdlIjo2fQ.G8MLWYagBJjT24Z1MDCA3HcY-RNvhGbwwP25Bai_2f4/X-FAB_Half-Year_Report_2024_ENG.pdf )
Think risks have been totally overblown on this name. China revenue was single figure in 2023, so mid teens in 2024? They will win the EV war so I don't see how they can compete without it. I don't see trade restrictions being focused on these lower value components. Tariff risk is real but in their Q3 letter Night Watch IM pointed out weakening FX helps to increase competitiveness of the German plants. Germany also reached 64% renewables in 2024, could get to 75% by 2026. Power prices only going one way - down.
Biggest threat looks like SiC and getting eaten by bigger fish, but that seems to be normalising.
Just the fact XFAB is receiving so many prepayments (not only from Melexis) shows the risk/reward on the expansion isn't quite what it seems on the surface.
Thanks for this - really helpful perspective that made me think!
You're right, I probably overstated some of these risks. That China number is a good example - the 55% growth sounds huge, but when you're starting from single digits, even that kind of growth keeps the total exposure pretty manageable.
Good catch on the German energy situation too. I hadn't realized renewables had hit 64% and where power prices were heading. Combined with the currency advantages you mentioned from Night Watch's analysis, the German manufacturing cost picture looks better than I painted it.
On prepayments - it is a telling signal. When customers put down €275.9M in deposits, they're essentially helping fund X-FAB's expansion. That does change the risk profile of their capex program.
On SiC - yeah, fair point about normalization. I might have focused too much on the current dip versus the longer-term story.
Really appreciate you taking the time to add this context. These kinds of discussions help build a better picture.
Nice writeup! For what its worth I was referring to the market's perception of risks rather than your piece. What's interesting is it seems to have derated down to essentially a cyclical Auto OEM/tier 2 supplier valuation at >8 PE, >5 EBIT as if trading into a downturn and profits will tumble 30-40%, while comparables are 50-200% higher. With the rest of the industry cutting capex, you could say the market is treating this expansion as a complete money furnace when they are still guiding for higher profits next year. So this resolves itself one way or the other anyway: expansion doesn't happen - minimal BS impact and shareholder returns, multiples improve. Expansion happens - shares rerate.
Thanks - this is really helpful framing on the valuation.
I take it you are quite bullish on X-FAB.
My understanding is that essentially the stock is priced for disaster but has two potential positive outcomes:
- Either they successfully expand (good)
- Or they don't expand and return capital (also good)
Correct?
Two questions if you don't mind:
1. Which companies do you look at as the most relevant comps? The peer group is tricky since X-FAB sits between pure foundries and auto suppliers.
2. As fleets transition from ICE to EV, does this generally mean more demand for X-FAB's type of analog/mixed-signal chips? Trying to understand if electrification is a tailwind for their core competencies.
Yes I mean if they genuinely see little RoI they will surely cancel what remaining capex they can. But I think potential problems are much more post expansion (scaling up/economics etc), because this is a structurally constrained market, as Covid showed.
1. From what I can see there are two groups of comps: the pure play mixed-signal/analog foundries (Vanguard Int Semi, Tower Semi, Hua Hong Semi, DB HiTek), and the larger integrated designers/manufacturers that have a focus on autos (the Infineons, SMTMicro, Renesas).
2. My impression is EVs require more of pretty much everything. There was another substack post on the subject, showing a slide from Melexis (the main customer). They are expecting 2-4x EV-related sales in 5 years. EV portion was 30% of their revenue in 2023, so coming off a substantial base. That's obviously moving the needle (and remember most of this will be outsourced to XFAB) https://open.substack.com/pub/siliconmatter/p/x-fab-an-overlooked-company?r=3swbr&selection=c42a4aeb-4cc2-4502-9cd3-13b08252b23c&utm_campaign=post-share-selection&utm_medium=web
Interesting they're most optimistic on e-braking, if that's already a large business at Melexis (don't know, haven't checked) that is very good news for XFAB.
This is being slammed from tariffs and rising euro. US exposure for XFAB/Melexis is 5-10%, another few % indirectly through EU customers. What are people's thoughts? I think orders can stay strong on China growth + UK being spared by Trump.
5-10% is too small to move the needle. This is a great buying opportunity for most quality companies that are not exposed to the US to a large degree.