[Fudzilla] TSMC outstrips Intel

[CHADBOGA]

Public scorn from a forum admin, so classy.

Why should rampant idiocy go unaddressed?


He got an infraction
Markfw900
Anandtech moderator

 

[SiliconWars]

This x1 billion.

I don't have any particular love for Intel. I don't have any emotional feelings about any corporation.

However, denying the fundamental supremacy of Intel as the titans in semiconductor is just silliness. They're going after mobile, and I would really not bet against them in the long run. Once they ramp up and put their full effort behind something, impressive results usually follow. Even mis-steps like Netburst and starting late on mobile haven't hurt them very much.

Their competitors would blush at the chance for half the profits Intel has. And R&D division, engineering talent, and fabrication technology? Forget about it.

Have you been living under a rock this month?

Bay Trail is struggling to beat S800 and Apple's A7...maybe you want to think more about who has the engineering talent these days.

While this article is junk, there are some important bits of info out there, one being that TSMC made $1.71 billion in profit in Q2. That's getting very close to Intels $2 billion Q2 profit. They are catching up in profit even though they only have about 1/3rd of the revenues of Intel. Cash is very important in this CAPEX intensive business.

As for your other points, Qualcomm already has more money, will make more profit in Q3 and never again fall behind Intel. Apple is already so far ahead in this regard it's basically unfair.

The process "advantage" Intel has does not seem able to give them better chips but still costs them a massive chunk of cash.

Not one company I mentioned in this post would rather be in Intels position - you need to get out of the Intel vs AMD thing and realise who the real threats are.

 

[liahos1]

Have you been living under a rock this month?

Bay Trail is struggling to beat S800 and Apple's A7...maybe you want to think more about who has the engineering talent these days.

While this article is junk, there are some important bits of info out there, one being that TSMC made $1.71 billion in profit in Q2. That's getting very close to Intels $2 billion Q2 profit. They are catching up in profit even though they only have about 1/3rd of the revenues of Intel. Cash is very important in this CAPEX intensive business.

As for your other points, Qualcomm already has more money, will make more profit in Q3 and never again fall behind Intel. Apple is already so far ahead in this regard it's basically unfair.

The process "advantage" Intel has does not seem able to give them better chips but still costs them a massive chunk of cash.

Not one company I mentioned in this post would rather be in Intels position - you need to get out of the Intel vs AMD thing and realise who the real threats are.

You miss the point. the gross margins on lagging edge nodes are vastly superior to the gross margins on leading edge nodes because of wind down of depreciation costs associated with older nodes. INTEL's gross margin AND profits from leading edge vs TSM's profits from leading edge is a more direct comparison. And there you'll see tsm isnt even in the same ballpark. Intel isnt in the business of building 180nm chips in high volume for the analogue players like Maxim or ADI. Just wanted to point this out to you. Since apples =! oranges.

 

[blackened23]

Why should rampant idiocy go unaddressed?

Yeah, that's what I was thinking. It's quite humorous seeing the live or die for AMD crowd salivating at such news, I just find it funny to see this happen in intel thread after intel thread over and over and over. There's no reason to even mention that you can't simply deduce profit levels from revenue, revenue != overall profit. None of these folks, apparently, know how to interpret an income statement. We can't have talk of that now, can we? let's just look at revenue and suddenly declare TSMC the winner when revenue has nothing to do with profitability - And intel makes more margins and profit per chip since they have control over their ecosystem. This is also obviously aside from the fact that Intel's technology is easily 3 years ahead of TSMC, because TSMC doesn't have FinFETs ready and will not have it ready before 2015. But by all means let's not let talk of reason interfere with the live or die for AMD folks who salivate at any chance they can to take potshots at intel. I don't understand it. While I myself don't like AMD CPUs, I don't really take non stop potshots at AMD all day either; I remain a fan of their GPUs and hope they return to their former Athlon glory days (even if it's a long shot). Do intel fans do the same in AMD threads? I don't know but intel threads tend to get dumped on all the time. Whatever, man. Let them have their "fun" I suppose.

 

[liahos1]

Yeah, that's what I was thinking. It's quite humorous seeing the live or die for AMD crowd salivating at such news, I just find it funny to see this happen in intel thread after intel thread over and over and over. Hilarious. There's no reason to even mention that you can't simply deduce profit levels from revenue, revenue != overall profit. None of these folks, apparently, know how to interpret an income statement. We can't have talk of that now, can we? let's just look at revenue and suddenly declare TSMC the winner when revenue has nothing to do with profitability - And intel makes more margins and profit per chip since they have control over their ecosystem. But by all means let's not let talk of reason interfere with the live or die for AMD folks who salivate at any chance they can to take potshots at intel. Whatever, man.

This is a good point ive noticed it reading these forums. Why do some people on these forums treat intel as if the company has murdered their first borns? Its unreal. People will twist, fabricate, subjectively interpret, and hold intel to a much higher standard than any arm based player or amd. It makes no sense.

 

[SiliconWars]

You miss the point. the gross margins on lagging edge nodes are vastly superior to the gross margins on leading edge nodes because of wind down of depreciation costs associated with older nodes. INTEL's gross margin AND profits from leading edge vs TSM's profits from leading edge is a more direct comparison. And there you'll see tsm isnt even in the same ballpark. Intel isnt in the business of building 180nm chips in high volume for the analogue players like Maxim or ADI. Just wanted to point this out to you. Since apples =! oranges.

Why is there so much obsession over gross margins? Why not profit-to-revenue ratio or some other marker that Intel is lagging in?

I'd rather be a company with $5 billion revenues and ~$2 billion profits than a company with $13 billion revenues and the same ~$2 billion profits. The company with the lower profit-to-revenue ratio has a much harder fall when things are going badly - and the situation just gets worse and worse for Intel with every passing quarter.

Yeah, that's what I was thinking. It's quite humorous seeing the live or die for AMD crowd salivating at such news, I just find it funny to see this happen in intel thread after intel thread over and over and over. There's no reason to even mention that you can't simply deduce profit levels from revenue, revenue != overall profit. None of these folks, apparently, know how to interpret an income statement. We can't have talk of that now, can we? let's just look at revenue and suddenly declare TSMC the winner when revenue has nothing to do with profitability - And intel makes more margins and profit per chip since they have control over their ecosystem. This is also obviously aside from the fact that Intel's technology is easily 3 years ahead of TSMC, because TSMC doesn't have FinFETs ready and will not have it ready before 2015. But by all means let's not let talk of reason interfere with the live or die for AMD folks who salivate at any chance they can to take potshots at intel. I don't understand it. While I myself don't like AMD CPUs, I don't really take non stop potshots at AMD all day either; I remain a fan of their GPUs and hope they return to their former Athlon glory days (even if it's a long shot). Do intel fans do the same in AMD threads? I don't know but intel threads tend to get dumped on all the time. Whatever, man. Let them have their "fun" I suppose.

Uhm, maybe you could point out where this is supposed to be happening in this thread?

This is a good point ive noticed it reading these forums. Why do some people on these forums treat intel as if the company has murdered their first borns? Its unreal. People will twist, fabricate, subjectively interpret, and hold intel to a much higher standard than any arm based player or amd. It makes no sense.

And you could as well? All I see is a couple of people who made a mistake by taking Fudzilla at face value, and then owned up to it. I sure can't see any "salivating" over anything but if all you have is a hammer then everything starts to look like a nail I guess.

 

[liahos1]

i see this in multiple threads. Anytime intel is possibly doing something good ppl downplay, obfuscate, subjectively interpret and are skeptical of it. People do the same to amd, mostly because the company hasnt really executed well in the past in its primary markets.

back to my earlier point. SiliconWars you miss the point of my earlier statements regarding TSM.

TSM is

1) smaller than intel including lagging edges
2) 5x smaller than intel at its leading edge (28nm) (taking INTC's COGS vs TSM's revenues into consideration)
3) Only generates a fab margin not an IEM margin (intel makes both)
4) makes the vast majority of its money fabbing for customers at lagging edges
5) You focus on Net Income which is a GAAP measure. Focusing on what really matters (free cash flow). You see Intel's free cash flow margin has been 50% to 100% that of TSMs. TSM has massive capex needs (same as intel) but doesnt generate nearly as much cash to cover them. In fact at the current trajectory the company is going to have to raise debt to cover its dividend after capex

one can argue all day long about intel's future prospects positive or negative. That doesn't change the above points though. You could be happier working at TSM for sure. But again doesn't change the following points where people are comparing apples and oranges.

 

[SiliconWars]

i see this in multiple threads. Anytime intel is possibly doing something good ppl downplay, obfuscate, subjectively interpret and are skeptical of it. People do the same to amd, mostly because the company hasnt really executed well in the past in its primary markets.

You do realise that both of those are your personal opinions and not facts?

back to my earlier point. SiliconWars you miss the point of my earlier statements regarding TSM.

TSM is

1) smaller than intel including lagging edges
2) 5x smaller than intel at its leading edge (28nm) (taking INTC's COGS vs TSM's revenues into consideration)
3) Only generates a fab margin not an IEM margin (intel makes both)
4) makes the vast majority of its money fabbing for customers at lagging edges
5) You focus on Net Income which is a GAAP measure. Focusing on what really matters (free cash flow). You see Intel's free cash flow margin has been 50% to 100% that of TSMs. TSM has massive capex needs (same as intel) but doesnt generate nearly as much cash to cover them. In fact at the current trajectory the company is going to have to raise debt to cover its dividend after capex

one can argue all day long about intel's future prospects positive or negative. That doesn't change the above points though. You could be happier working at TSM for sure. But again doesn't change the following points where people are comparing apples and oranges.

The points you have missed are -

TSMC is growing much faster than Intel.
TSMC's customers are growing much faster than Intel.
TSMC's customers have much more money, free cash flow etc than Intel.

It's these customers that are fueling TSMC's progress because they are relying on TSMC's manufacturing.

Also missed is that Intels 22nm does not appear to be any better than TSMC's 28nm in terms of what happens when the chips pop out at the end. If it were, Bay Trail would have been a "slam dunk" and instead it's barely competitive with the better ARM offerings. Die sizes appear to be no smaller than the competition either.

So Intel pays the "leading edge" CAPEX penalty for what basically amounts to zero competitive advantage. It's fine to look at Intel and TSMC right now and say "Intel is much bigger", but the gap is shrinking, the spending on CAPEX is now almost equal and the companies that are bankrolling TSMC's progress have much more financial clout than Intel will ever have. You need to look at how much the gap is shrinking, and how fast.

 

[MisterMac]

Public scorn from a forum admin, so classy.

Sucks when you can't live in the dream-world?

Classy.

 

[liahos1]

"TSMC is growing much faster than Intel."

Sure no arguing that. TSM has grown 69% from the cyclical semi bottom in 2008 and Intel has grown 41%. Although on total dollars intel has added 15.7bln dollars to its revenues while TSM has added 6.9bln dollars.

"TSMC's customers are growing much faster than Intel."

Not all of them. More than half of TSM's biz is fabbing for GDP growth type semi companies which havent really grown revenues consistently over the past few years. pure cyclical commodity semi companies. The other half is mobile which has been growing explosively.

"TSMC's customers have much more money, free cash flow etc than Intel."

Who cares? Those customers are also shifting business away from TSM as the company has lost pricing power by losing 28nm monopoly status. QCOM, Mediatek etc are all moving business to glofo and samsung. Thats common knowledge. Besides those customers are taking that cash and not building fabs for TSM. So your point is invalid.


"It's these customers that are fueling TSMC's progress because they are relying on TSMC's manufacturing."

TSM again is just one of a few manufacturers now at 28nm. Samsung seems to be aggressively stepping up its 20nm efforts. Intel is poaching TSM customers as well. I expect as intel increases total die capacity at 450mm and 10nm if it cant fill that capacity fully with their own chips they will continue to assault TSM's base of business by poaching customers who would be more than happy to fab at intel.

Your point about intel deriving no value from being at the leading edge is silly. Again i would ask you to look at how much they make at the leading edge vs TSM. You can't obfuscate simple arithmetic. Their ROIC is much higher than TSMs. This is a known fact!

 

[dahorns]

You do realise that both of those are your personal opinions and not facts?

The points you have missed are -

TSMC is growing much faster than Intel.
TSMC's customers are growing much faster than Intel.
TSMC's customers have much more money, free cash flow etc than Intel.

It's these customers that are fueling TSMC's progress because they are relying on TSMC's manufacturing.

Also missed is that Intels 22nm does not appear to be any better than TSMC's 28nm in terms of what happens when the chips pop out at the end. If it were, Bay Trail would have been a "slam dunk" and instead it's barely competitive with the better ARM offerings. Die sizes appear to be no smaller than the competition either.

So Intel pays the "leading edge" CAPEX penalty for what basically amounts to zero competitive advantage. It's fine to look at Intel and TSMC right now and say "Intel is much bigger", but the gap is shrinking, the spending on CAPEX is now almost equal and the companies that are bankrolling TSMC's progress have much more financial clout than Intel will ever have. You need to look at how much the gap is shrinking, and how fast.

Saying that BT is barely competitive with the ARM offerings is more than a little intellectually dishonest. BT, core for core, tops all ARM offerings except the A7. GPU is better in the ARM offerings, however we really haven't seen a solid comparison in terms of power usage. Indications are that BT is very power efficient. If it uses less power to reach even "barely competitive" performance, then I'd say that is a win for Intel. You also leave out top line performers, such as Haswell, where Intel has no equal.

In addition, to say that CAPEX is almost equal, last I checked, is only true if aggregate the CAPEX of the entire ARM eco-system. Unless they all work together (they don't), that comparison is pointless.

 

[liahos1]

Saying that BT is barely competitive with the ARM offerings is more than a little intellectually dishonest. BT, core for core, tops all ARM offerings except the A7. GPU is better in the ARM offerings, however we really haven't seen a solid comparison in terms of power usage. Indications are that BT is very power efficient. If it uses less power to reach even "barely competitive" performance, then I'd say that is a win for Intel. You also leave out top line performers, such as Haswell, where Intel has no equal.

In addition, to say that CAPEX is almost equal, last I checked, is only true if aggregate the CAPEX of the entire ARM eco-system. Unless they all work together (they don't), that comparison is pointless.

It even tops the a7 (albeit less so). Given that the scores we've seen are all on benchmarks nobody seems to care about (when intel is shown to be leading) but ppl love when anything arm based seems to be doing well, i find it amazing that baytrail does better on windows based scores and still manages to outperform on some android vs ios based scores.

 

[SiliconWars]

Who cares? Those customers are also shifting business away from TSM as the company has lost pricing power by losing 28nm monopoly status. QCOM, Mediatek etc are all moving business to glofo and samsung. Thats common knowledge. Besides those customers are taking that cash and not building fabs for TSM. So your point is invalid.

It doesn't matter who is fabbing the chips, all that matters is that somebody is. All Apple and Qualcomm needs is a fab partner that can offer them something close to what Intel's leading edge node is. Be that TSMC or GF, it simply doesn't matter.

TSMC will be on 20nm before the rest and will have a similar position to what they enjoy now. After that who knows what will happen.

TSM again is just one of a few manufacturers now at 28nm. Samsung seems to be aggressively stepping up its 20nm efforts. Intel is poaching TSM customers as well. I expect as intel increases total die capacity at 450mm and 10nm if it cant fill that capacity fully with their own chips they will continue to assault TSM's base of business by poaching customers who would be more than happy to fab at intel.

Who? Small time business that isn't going to make any real difference. Intel has gone on record many times to say that they will not enable their competitors. That's why Apple is at TSMC and why they'll never be at Intel.

Your point about intel deriving no value from being at the leading edge is silly. Again i would ask you to look at how much they make at the leading edge vs TSM. You can't obfuscate simple arithmetic. Their ROIC is much higher than TSMs. This is a known fact!

The vast majority of Intel's profits come from servers where they have no competition - yet. Who do you think will be fabbing those competing chips?

 

[SiliconWars]

In addition, to say that CAPEX is almost equal, last I checked, is only true if aggregate the CAPEX of the entire ARM eco-system. Unless they all work together (they don't), that comparison is pointless.

TSMC's CAPEX is basically equal to Intel's now, the ARM crowd combined will be far ahead.

 

[CHADBOGA]

TSMC's CAPEX is basically equal to Intel's now, the ARM crowd combined will be far ahead.

The ARM crowd combined are wasting billions on duplication, so hardly an apples to apples comparison.

 

[Markfw]

Note to everyone here, Imouto got an infraction for his mod callout. This is not tolerated here.

 

[Phynaz]

you need to get out of the Intel vs AMD thing and realise who the real threats are.

You are the first person to mention AMD in this thread.

Feeling defensive?

 

[SiliconWars]

You are the first person to mention AMD in this thread.

Feeling defensive?

Who else can he be meaning? Arkaign keeps repeating the same lines about Intel's overwhelming "fundamental superiority" in all things - clearly he can't believe that includes Apple and Qualcomm and Samsung who have more money, market share and by the looks of things better engineers as well?

 

[krumme]

TSMC's CAPEX is basically equal to Intel's now

And with lower wages. They are in a very steep learning process. Armv8 is comming. The mobile market is still exploding. The arm ecosystem rocks and gives perfect depreciation. Their business model is perfect and the consumers go their way. Apple just showed the huge potential of armv8. I think tsmc is in a perfect situation. I cant see any new threats - anyone?

As a general note i think we often tend to look and discuss to narrow on the company (strenghts weaknesses) and not the context (opportunity threats).

 

[aigomorla]

how is AMD even included if they dont own fabs?

And personally im not suprised TSMC finally outplaced intel.

Lets think realistically what TSMC does and who they supply for.
Intel on the other hand only did Intel.
I think just earlier this year they contemplated on renting out there retired fabs to other people, however intel has always been quite snobby about what they make, everything that comes out of their fabs were stamp'd INTEL.

To me TSMC out ranking intel tells me thats how large of a technology explosion we had in smart phones + tablets in the past couple years.

clearly he can't believe that includes Apple and Qualcomm and Samsung who have more money, market share and by the looks of things better engineers as well?

They only have better engineers because they are not in the desktop/cpu sector... and intel is trying to use desktop/cpu engineers inside the smartphone arena... which yeah.. in that definition doesn't work right.

Bring those engineers on AMD and see if they can compete with intel...
They will sink that boat so fast, you wont even have time to swim to a life raft.

 

[kevinsbane]

And personally im not suprised TSMC finally outplaced intel.

They haven't outpaced Intel. Under no objective measure have they done anything of that sort.

Except in the Fudzilla's numbers, that is, which have no resemblance to reality (as evidence by this post and this post)

 

[liahos1]

apparently some people still haven't looked at the numbers. Numbers speak a very direct language that i guess some people dont want to read, lest it destroy the fabric of their alternative narratives.

 

[dahorns]

TSMC's CAPEX is basically equal to Intel's now, the ARM crowd combined will be far ahead.

1. That isn't true - TSMC CAPEX for the year is estimated at $9.5-$10 billion. Intel is shooting for $11 Billion.

2. You realize that CAPEX isn't the same as R&D, right?

Intel = http://ycharts.com/companies/INTC/r_and_d_expense

TSMC = https://ycharts.com/companies/TSM/r_and_d_expense

Intel spends over $2 billion a quarter. TSMC spends less than $.5 Billion.

 

[SiliconWars]

1. That isn't true - TSMC CAPEX for the year is estimated at $9.5-$10 billion. Intel is shooting for $11 Billion.

2. You realize that CAPEX isn't the same as R&D, right?

Intel = http://ycharts.com/companies/INTC/r_and_d_expense

TSMC = https://ycharts.com/companies/TSM/r_and_d_expense

Intel spends over $2 billion a quarter. TSMC spends less than $.5 Billion.

1) I didn't mention R&D anywhere.
2) Intel has cut CAPEX twice this year already while TSMC has raised theirs.

 

[dahorns]

1) I didn't mention R&D anywhere.
2) Intel has cut CAPEX twice this year already while TSMC has raised theirs.

Yes, they cut it to 11 Billion. I know you didn't mention R&D, but I fail to see why having a larger CAPEX expense is significant of anything. Companies build their fabs at different times. Who cares if Samsung spends more on CAPEX in one year than Intel? all that means is that Samsung is building a Fab that year and Intel isn't. It certainly doesn't indicate that Samsung is catching up technologically with Intel.

For example, they may be building a fab that is 2 nodes behind Intel. They incur the expense in order to stay within striking decent or to keep up with production, but they aren't "catching up".