AMD is now worth more than Intel
seekingalpha.com
seekingalpha.com
From Haswell (4000-series) to the 12000 series I would say there was maybe one generation of improvement in between and six marketing generations where nothing got better nothing got faster but prices were outrageous -$600 for a slow hot MacBook cpu, and everything got new marketing names!
So the industry has been desperate for an alternative to Intel since 2009 and AMD has finally delivered on the promise by building data center chips and adapting them to laptop use. I was an early AMD investor because of rx480 at $3.50 by the way but I sold out at $10.50.
From here on AMD will truly have to earn its keep and things will be a lot tougher as they are no longer the scrappy underdog anymore!
Good news for consumers either way though. Intel already have their crown back in the midrange at least.
Toss in global politics with TSMC and Intel looks like a much safer bet long term. What makes AMD more attractive from an investment point of view? There's got to be some reasoning for it and I'd love to understand it.
Imho paying a dividend is a pretty bad sign in this industry - they should be funneling everything in to R&D. Their 10x P/E reflects a belief that their earnings are going to remain relatively stagnant.
See these forward looking earnings estimates:
https://www.marketbeat.com/stocks/NASDAQ/INTC/earnings/
https://www.marketbeat.com/stocks/NASDAQ/AMD/earnings/
That said, I'm with you, I'd rather be holding Intel in this market.
Unlike most meme stocks however AMD is a good business regardless
Right now the bet, or at least the market suggest, Intel wont be able to catch up to TSMC in terms of technological leads, and even if they do, they wont be able to do it as cheaply as TSMC. x86 is being eroded by ARM, and even ARM are not safe from the Riscy Silver Bullet.
The strange thing about the market in the past 20-30 years is how they view manufacturing and machinery as liability. For Intel without the technological moat ( in terms of leading edge ), Fabs are not viewed favourably by the market. Does it makes sense, of course not. But the the markets can remain irrational longer than you can remain solvent.
Intel is advancing their work far quicker than I could expect. While catching up to TSMC is still a tall order, at least they are finally innovating. Instead of saying No to their Custom Foundry customers they are now being challenged to deliver exactly what their customer demands.
Also, it’s a massive risk.
If, for whatever reason, there is a drop in chip demand (which is entirely possible with a global recession, which is looking increasingly likely) Intel will be burning billions keeping their manufacturing running.
AMD, on the other hand, simply need to not place any more orders and offload the cost to TSMC and GlobalFoundries, etc.
Also, dividends probably impact PE negatively. Big investors invest money in companies because they have done their research and believe those companies will do a better job with their money than other companies. The company giving them a dividend back simply means they need to figure out something else to do with that money, which logically should be to reinvest it in the same company (since it’s the one their research says will currently meet their objectives best). The dividend has simply added a lot of operational costs, and in many cases and in many cases taxes, for something that is completely unnecessary for them.
Further, I suspect Intel is increasingly tying itself to the US, whereas AMD, while still heavily based in the U.S., isn’t doing that as much. If we do see a global technological schism, AMD will be able to straddle both sides in a way that Intel cannot.
Stepping back, AMD has process advantage in x86 via TSMC and is executing very well. Clearly major opportunity to grow with low risk at Intel’s expense.
Meanwhile Intel has to implement some big changes to stay competitive, has major capex needs and is trying to break into markets where it isn’t present (or competitive) which is clearly risky.
In any event no one is saying anything is assured only that some things are more likely or risky than others. Intel has a lot to do and recent execution has been weak. You may think that will change but the market will need evidence. AMD on the other hand is in the opposite position and gets credit for that.
Also you may think it’s laughable but there is no should about it - it’s a statement of fact that AMD’s market cap is more than Intel’s and that fact also plays to AMD’s advantage in some ways (acquisitions for example).
Profits/capital
I don't think this is necessarily true. AMD doesn't seem to do well beyond 35W TDP [0]. And for such "mobile" workloads we have a tough competition coming from ARM (specifically from Qualcomm's Nuvia team) by the end of this year [1]
[0]- https://www.anandtech.com/show/17276/amd-ryzen-9-6900hs-remb... [1] - https://www.anandtech.com/show/17075/qualcomm-x-nuvia-silico...
To be clear, I don't think that the investment case rests on AMD outperforming Intel everywhere. They just have to be competitive for long enough in enough segments to take material market share and I think there is a very credible case for that happening.
Edit: Curious if the downvotes are because you think AMD doesn't need TSMC, because you think China will absolutely never invade Taiwan, or if you think AMD can find another fab within a year or two.
The US _did_ have a mutual defense treaty with Taiwan, but it was terminated by the Taiwan Relations Act after the US initiated diplomatic relations with the People's Republic of China on Jan 1st, 1979.
China is very important but the shipping lanes to SE asia and regional stability is more valuable. You can sanction China and get in a sanction war with them until one side gives in. Or a trade war like trump tried.
Their focus however is to win the war before it starts unlike the US which is to win the war after it starts (imo).
If it gets to a full in sanction like russia, the western economy will tak a massive hit, but it isn' beyond recovery and might actually be good to force resilent supply chains. Especially with silicon fab.
How many years do you think it would take to build up those resilient supply chains? 5 years?
That's 5 years without most products from AMD, Nvidia, Tesla and every other TSMC customer. Samsung can absorb some of that business, but not much. Intel will shoot to the moon in that scenario.
I can't tell you what will happen if Taiwan is invaded tomorrow, next year, or twenty years from now, but I can tell you that the US will think long and hard about whether or not it will want to get involved. And it might well choose not to.
What planned or potential pipeline of other new non-Intel chip fabrication plants on US soil do you foresee?
The TSMC fab in Phoenix is build for 5nm and will open in 2024. The TSMC 3nm process should go into mass production in 2023, it is expected that the 2023 iPhone will use a 3nm SoC. The fab will only produce 20k Wafers per month, the TSMC gigafabs in Taiwan produce 100k Wafer per month each.
Details about TSMC Phoenix fab: https://www.cnbc.com/2021/10/16/tsmc-taiwanese-chipmaker-ram...
Thanks for your assessment on this. Would you please share how you arrived at one year of a product sales gap?
Do you think it is possible for the US to bring 3nm process to the US while that is still relevant?
(1) TSMC is building fabs in both the US (Arizona) and Japan (Kumamoto). (2) China is unlikely to invade Taiwan anytime soon; the country is working on more of a diplomatic takeover a la Hong Kong. (3) I don't know any reasonable argument for AMD not needing TSMC.
The historical circumstances of HK and Taiwan are nothing alike. Furthermore, the mainland communist party doesn't have any sway on Taiwanese politics, as far as I can tell ...other than as a common nemesis to unite against in relation to any issues that could affect the island's autonomy.
Closer coordination and cooperation in a socioeconomic context? Also on the table.
But _integration_ with the communist PRC? Uhmm... yeah no, that's a pretty disingenuous characterization of the KMT side of Taiwanese politics.
China wants more of these people and are actively employing soft power in Taiwan to convert more to their side. I'm not sure what the intended time frame for success is, but China is in this for the long-haul.
Perhaps if things goes China's way, the PRC can strike a deal with the KMT for some sort of local governance, likely along with a name and flag change - or a third party could rise up and supplant KMT. Of course, either situation would kill off the DPP.
No, what I find highly implausible is the notion that such an extreme "unification at any costs" ideology has any significant popular support. More so still in the aftermath of the PRC's recent suppressive tactics in Hong Kong.
I'm sure the scenarios you hint at are indeed explored at great lengths among Beijing-funded think tank strategists, but I can't see Xi & co being naive enough to elevate such an approach as their most viable course of action.
https://www.worldstopexports.com/taiwans-top-import-partners...
(I believe even South Korea and Japan export more to PRC than they do to USA)
https://www.worldstopexports.com/taiwans-top-import-partners...
(I believe even South Korea and Japan export more to PRC than they do to USA)
What is your point?
For context, AMD's quarterly revenue is ~5B and Intel's is ~$20B.
AMD's quarterly net income is ~$1B and Intel's is ~$4.5B.
I'm surprised that AMD is valued higher, despite their revenue and profits being so much smaller. The financial markets are in essence betting that AMD will be able to achieve revenue/profit parity with Intel, despite the huge lead Intel currently has. It will be very interesting to see if that comes to pass.
I remember the days when AMD seemed to be on the verge of bankruptcy, and there was talk of Intel simply buying AMD since it was so cheap. Presumably that didn't happen because Intel feared anti-trust laws if it purchased its main/only competitor. Kudos to AMD for the turnaround - this competition is sure to be benefit the wider industry.
AMD used to run its own fab and nearly went bankrupt doing so. It spun off the fab as Global Foundries, and still couldn't catch up due to contractual obligations to use GF which was behind Intel. By eventually switching to TSMC (with new/renegotiated GF contracts; GF still makes glue chips on chiplets and also got some AMD stock), AMD beat Intel with more advanced process (it was already competitive in architecture).
Intel is now the one that's disadvantaged by using less than state-of-the-art fabs, the position AMD was in until about 5-7 years ago. Not only is Intel disadvantaged, it's doubling down on fabs, imagining that it will somehow overtake TSMC in the future.
In 2022, Intel is going to have its first negative free cash flow in at least 15 years. It's going to have negligible free cash flow for the two years after. All of its cash flow and then some is going to build out fab capacity and node tech as it tries to dramatically increase capacity for its design business and its foundry business. Meanwhile, their most lucrative business segments, in particular datacenter, are getting gobbled up quickly by the competition. In Q1, Intel decreased -5% in cloud YOY despite that segment growing quickly.
If you believe that everything will go well with Intel, then this is money well spent and the stock is cheap. But given their historic difficulties and the quality and quantity of their competition in XPU design and manufacturing, there's a meaningful chance that their profitability will be greatly diminished.
Correlation. Not causation.
Apple shied away form Intel due to their failure at landing mobile chips.
Intel failed at mobile due to their (then) sales-focused (rather than engineer-focused) regimen.
Source: I worked at intel on the Mac chips for a year and a half
iPhone is largely what created the sustained, predictable volumes for high-end chips that has allowed TSMC flourish, and TSMC is what has allowed AMD's strategy to take flight.
If AMD was stuck with GlobalFoundry, they would be nowhere.
In five years or so (ok maybe 7 or 8) Intel turned a two year lead in fab technology to being behind by two years, so there's that too.
Utter inability to break into the mobile market, ARM or otherwise.
Failed attempts at discrete graphics and not having something ready for the rise of crypto/AI.
XPoint was basically a failuer, SSD dominance taken from them.
It shows both an empty cupboard of in house talent and technology and a zombie management structure.
DCF using ^TNX as discount rate: $70.93, with a 50% discount: $35.465. Current price as of Mar 5, 2022: $108.41.
BTW, not only did you respond with smugness and a narcissistic sense of superiority but you ignore my question, which was an opportunity to display your expertise, were you to have any. You are out of your element with financial talk. Stick to programming.
If you did discounted cash flow analysis on Facebook or Google, you'd find that despite their high price to earnings ratios, their earnings produced fair valuations, not overvalued ones.
Understanding the future cashflows of a business is really difficult stuff. Playing with numbers in a spreadsheet is really easy stuff.
If you’re paying $42.14 per dollar of earnings there is a likelihood that more often than not you’re overpaying.
If you can’t pass simple mechanical litmus tests you’d better be damn sure, and frankly there just aren’t many exceptional companies where I feel damn sure about them.
You can't just do a simplistic extrapolation of the last couple years in a spreadsheet.
Google couldn't pass your "simple mechanical litmus tests" when it went public. Apple couldn't pass it in the late 90s. Amazon couldn't pass it most of it's life. IBM could for a good number of years (just look at the numbers when Buffett was buying it...) Most of the banks could in 2005/6/7.
You seem to think that I use these mechanical estimates as the primary method for purchase, but they are not.
I'm not sure what your point is about IPOs: they're historically a terrible time to invest, which is common knowledge.
And actually all of the companies you have mentioned have easily passed such litmus tests over their life time at different times. Sometimes they're overbought, some times they're oversold. If you buy when they're overbought, you'll end up with a lower IRR than if you hadn't.
I didn't make a point about IPOs. I made a point about companies where real insight was required, both to see good investments and not make bad ones. Simple rules like "IPOs are bad", "dont pay x times earnings" etc, they haven't worked in years.
If it's all as simple as you say - really - just go start a hedge fund. You can start an LLC in a day, throw some money into a brokerage account and be up and running. There are still plenty of LPs who will invest in a simple long only fund, even with vol, as long as the performance is good.
Correct about what? What people are willing to pay for the company at that point in time? Sure. What the value of discounted cash flows generated by a company over a span of years? Definitely not.
What warrants such a huge drop if the market was correct about FB? Seems to me like it was indeed overvalued, but the market just came to that realization a few months ago. Even though we're experiencing a broad selloff in the market FB has been hit way harder than comparable companies.
I think the issue is that AMD's growth seems like it might hit significant headway. Samsung is having issues with their fabrication. We'd seen articles about AMD and Qualcomm wanting to move a lot of their chips to Samsung because TSMC was seen as too much of an Apple shop to their detriment. We've seen articles in the past month noting that Qualcomm is moving things back to TSMC from Samsung because Samsung's yields are so bad. The point here isn't that the sky is falling or anything, but that AMD does face some manufacturing constraints.
We've seen reports that Intel has secured a large amount of TSMC's 3nm production and it's safe to assume Apple has as well. Qualcomm will also be competing for fab space. In that environment, will AMD be able to keep pushing its advantages?
AMD has done great stuff over the past few years. Some of that has been helped by Intel's fab mishaps and TSMC's wonderful advances. That's not to downplay the many other things they've worked hard on. Many companies get openings and never take advantage of them. However, I think it's important to note that the opening is likely going to be narrowing.
If Intel is able to keep AMD away from TSMC's 3nm fabrication long enough for Intel's own fabs to get back to being competitive, what does that mean for AMD? Let's say that Intel 3 launches in mid/late 2023 and it's truly equal to TSMC 3nm. Let's say that Intel keeps AMD away from TSMC 3nm until 2024. What does that do to AMD's growth?
AMD is looking to launch TSMC 5nm Zen 4 processors in the second half of 2022. We've seen reports that TSMC is going to be making 3nm Intel hardware from July 2022. Do we see AMD 5nm processors in September and 3nm Intel processors in December?
Again, I don't want to take anything away from the wonderful work AMD has done. At the same time, I think most people would likely agree that with equivalent fabrication, Intel is as good as AMD if not better. If Intel is able to launch 3nm and box AMD out of 3nm for a while, that can offer Intel a while to leapfrog AMD on process - and possibly enough time to make sure that Intel's fabs get back in the game.
Yes, AMD had great growth in 2021 and they'll have some great growth in 2022. 2023? It seems like Intel might leapfrog AMD there.
Again, the sky isn't falling and I believe that AMD will have a lot of success in the future. It's more that it seems like AMD's growth will get stymied by Intel's use of TSMC and Intel's own fabs catching up. The idea behind AMD's current valuation is that they will become as big as Intel and even a lot bigger over the next several years. However, even with amazing growth in 2022, they won't become as big as Intel. In 2023, it seems like a lot of AMD's advantages turn into Intel advantages as Intel launches on TSMC's 3nm process. 2023 seems like it'll be a great growth year for Intel - at the expense of AMD's growth. I think AMD is in a strong position to continue being a great company with great products. I just think their growth slows as Intel fixes its problems. Of course, this all assumes that Intel fixes its problems.
FWIW: I want competition for the sake of progress, but I am bullish AMD for the next few years, and may be bullish Intel once they demonstrate their ability to shrink their process and execute IDM 2. Then they compete with AMD again… and TSM.
Intel had the greedy stupid arrogant policy of going it alone for everything, and failed to onboard cellular (3x!) and AI and flash and tv and the list goes on and on and on .... Hence they wasted more than 10 years without onboarding a single new type of semiconductor into the Intel Fab production process! It doesn't matter if they designed the best chips in the Universe they can't keep up if they can't build the next generation Fab and they can't do it with CPUs alone so at the moment they are facing total financial destruction of in house CPU production!
Taiwan Semiconductor had the wonderful intelligent smart idea to crowdsource designs from all over the planet and of course that open market for foundry services is the only way forward and Intel was a fool for a decade at least! It's still not clear they can mend their ways!
Unless Intel proves it can succeed as a foundry, they're doomed - or will lose their fabs! It's not an easy transition to make for such a large company!
You seem to think it's a minor thing for Intel to pick itself up and start driving again. No. False. The other competitors have boarded jet airplanes (foundry model with much greater economies of scale) whereas intel still thinks it can compete in the cars (in-house fabs) of 50Y ago ..
Along with Tesla, AMD is a "won't go down but probably won't recover fully after a big crash" type of valuation.
An average first year finance student can do a "brief DCF analysis". They put garbage in for the cashflows and get garbage out for the valuation - and then write it up with high precision.
andrewmcwatters.com boasts a $1400/day rate. That's not billions, but is still quite a respectable rate for their services. I assume they're well respected and trusted by their clients.
Hoever, in your comment here, you've mistaken some napkin math for a hard and fast assertion on their part and I think it's a little unfair. There's far more to a company's existence than DCF, all the poster did was point that out. It's obviously not that easy to run a hedge fund, but that's no excuse for sloppy facts here, or anywhere else for that matter.
Napkin math of "I think they can sell x units at y margin with z fixed costs because of a, b and c, and I think that will grow at g because of whatever and my back of the envelope says that is worth $" would be great. Ie, real insight. Low precision, high accuracy insight with napkin math would have been wonderful.
Whenever I've tried to run my own DCF analyses, I've found some parameters and modelling assumptions have a major impact on the resulting estimate. Discount rate parameter is one. The model and assumptions used to estimate revenue and operating/net profit when extrapolating 5 - 10 - 20 years into the future also make a large impact.
Suppose in our DCF analysis we aspire to forecast company performance over a 25 year time horizon and then discount the profits back to some NPV. Then we need some way of extrapolating the company's revenue for the next 25 years. One way to do this could be to put together some kind causal model where we identify the major factors that contribute toward generating revenue -- maybe this could break down revenue by segment or major product line, and attempt to predict supply, demand, price points, production capacity etc for each product line, using knowledge of how the industry works. Then we'd have to figure out how to forecast all of those drivers of revenue, e.g. forecasting the capex used to build new plant to increase capacity & so on. Another way could be to ignore trying to build a semi-plausible generative model for revenue and instead just do some kind of simple empirical model only -- e.g. fit a regression model to the trailing 5 years of historical revenue numbers or annual revenue growth rates combined with some kind of prior to force revenue growth to decay toward something unobjectionable the further into the future we project.
I've got some code that automates calculating a DCF. I've adopted some form of the latter approach to forecasting revenue -- fit a simple empirical statistical model to historical revenue -- it isn't very intellectually satisfying as it doesn't incorporate any real world knowledge about the causes of revenue in specific industries or companies. So, many forecasts of revenue generated by this approach will be quite wrong, leading to quite misleading DCF estimates. But on the other hand, I can run my kind of dumb DCF analysis in a completely automated way using inputs that are reported in company financials in a standard format, which is much quicker and easier.
When i run my crude DCF analysis over INTC and AMD -- using a somewhat arbitrary 6% discount rate -- I get the estimates
INTC -- priced by mr market at $48.07 / share
quantile 0.2 $87 / share
quantile 0.5 $140 / share
quantile 0.8 $196 / share
quantile 0.2 $42 / share
quantile 0.5 $77 / share
quantile 0.8 $121 / share
market valuation for AMD is either based on something that isn't discounted future earnings, or market's forecast for how future earnings will evolve is very different to my crude backward-looking model.
I use the 10-year Treasury note as a risk-free rate, which results in a share price much too high due to historically low interest rates. I do not have a reasonable risk premium that I otherwise use, as I have not yet found a sensible contemporary measure of such a premium. I do know that investing with a desired 10% real return is a relevant figure, but I don't have many other data points on that. As a result, I use a 50% discount from the resulting price to provide a margin of safety.
These are all roughly textbook constants that I use to serve as some mechanical "suggested basis [for] maximum appraisal for investment."
I also perform these in an automated way with publicly available SEC data... I don't believe that it is supposed to be intellectually stimulating. I think it's boring, and I'm fine with that. There is no simple mechanical process that will allow one to appropriately incorporate trade information and adjust valuations as a result. You don't know until you see the reports. And most of the time, not really even then. I don't know what the BOM is for plenty of products. I don't need to research that.
I don't enjoy analyzing companies that have too much at stake to produce their net earnings; even companies I personally enjoy I have a hard time convincing myself even with a fair value estimate that I'd like to own its common stock.
I do appreciate your reply, though. I'm more interested in these discussions than ones simply claiming how supposedly great or immoral or how much potential a company has. If I want opinions to fuel speculative theories, I can go to Twitter.
Think of putting in the last 10 years for blockbuster, or newspapers, while in the late 90s. No good.
trying to make forward-looking predictions extrapolating a decade or two out based upon backward-looking financial metrics may only make sense in situations where you have reason to believe that the conditions and dynamics that generated the past results will continue to persist for the next decade or two into the future (and it might not even make sense then)
i invested in one company late last year based on an initial screen using this kind of extrapolation of historical financials, and a bit more reading through the company's annual reports to get a better feel of the business model. the business model seemed a little weird but the business' financials looked good, especially compared to the company's share price.
since then i've updated my expectations and expect to earn about -80% to -100% return on invested capital. i failed to read symptoms of possible accounting fraud that were signaled in public disclosures on the company website at the time i invested, and a few months later directors have resigned and been replaced, the company has been delisted for failing to file quarterly financials, and the company has released the results of an accounting probe that has explained how the former directors fabricated all of the company financials, starting off with dating sales before goods were delivered to customers, and ending with making payments to the company from their personal accounts and registering the payments as if they were sales from external customers.
part of operationalising this kind of valuation model into a profitable investment scheme also needs a supporting due diligence function, which in this instance i failed dismally at
To add to that: If country A depends on products of country B, then country C could want to attack country B to hurt country A.