Many seem to be recommending AMD now, as it’s hit this lofty level… Does this still look like a time to buy if you’re not already in? or to increase one’s holdings? I am just finding it a bit strange how the raised targets and recommendations etc. are showing up right now…
I saw a speaker on Schwab tv who was saying that in the past when AMD was doing well, they usually under estimated earnings. HE suggested that what if the AI market for AMD is better than anyone is thinking. What if AMD makes 3 or 4 billion on AI instead of the 2 that Lisa has said? Then he went on to talk about how AMD is making the superior CPU’s (especially server), how they make gpu’s that are competitive, how they make a variety of other products like for cars and phones and make the point that they are more than a gpu maker. Next there was an article from TSM that they expect 20% growth instead of 10% in 2024. Then nvidia released a report that the AI market is still hot too. Add that all up and maybe thats why AMD has popped above it old all time high from 3 years ago, but that is just my guess…doc
AMD doesn’t make sense to me. NVDA has the market and AMD and Intel are picking up the scraps. If you look at NVDA it is cheaper and analyst expect this cycle to continue for at least 2 years so why buy AMD. Just go with the clear winner. At least that is my thoughts.
It’s just hard to explain Andy. Some of us have been here talking about AMD since the 90’s when the founder of AMD Jerry Sanders was a regular poster on the AMD thread. He left because of SEC rules. My first computer had an AMD K5 cpu with a math coprocessor just to let you know how long some of us have been here. I agree with you that NVDA is the beast. Back in those days, Intel was the 800 pound gorilla. AMD has a special place for some of us. I believe that nvidia will go to $1000 soon like next 12-24 months but i haven’t been able to pull the trigger on NVDA. Maybe I should…doc
edit: Of course I also think that AMD will get to $200 maybe this year.
I was not here when Jerry was contributing himself, certainly, but I’ve been here for decades. Emotional responses around AMD have cost me a lot of money over the years, both in times when the company was suffering and I couldn’t bring myself to sell, and in terms of missed opportunities like investing in Nvidia sooner. But they have also made me a lot of money. Right now I have exposure to both.
AMD has increasingly competitive solutions that will continue to grow share in a rapidly growing market, that’s their opportunity. Nvidia is essentially defining and expanding that market, that’s their opportunity. Until and unless I see Nvidia taking serious server CPU share or personal desktop CPU share, I’m not worried about AMD at all. All. They’ve clearly mastered execution in a way that they have never had prior to last seven or so years. They’re going to be incredibly hard to kill, in fact, I don’t think anyone can do it.
I think there’s room for everyone. Intel is the one I’m watching with Very curious eyes. The foundry business versus everything else they do is going to be strange as it plays out as more than just an enabler for the CPU and GPU markets.
I get it Doc and I see many great investors that have AMD in their portfolio. But for a play on AI , when NVDA can’t supply enough chips, it just seems that AMD is second best. Put them both on a chart in any time period from 1 year on out and see what I see. Recently AMD has been doing better but I do not expect that to sustain itself. I am not against AMD but I just think they are second fiddle to NVDA so I would rather just keep NVDA and put anything I have for AMD into NVDA.
I would agree except the foundries are having a hard time just keeping up with NVDA.
I suspect the reason AMD and Intel are focusing on AI is they have seen how much Nvidia can charge for AI hardware (and with fantastic margins). AMD was not really late to the race, they were busy with the MI300A for the El Capitan supercomputer. It’s interesting that they repurposed that to the MI300X for AI applications. It’s AI that’s driving the market and Nvidia’s margins. The market for specialised hardware is expected to increase hugely in 2024 and thereafter. Why shouldn’t AMD try to attract some business to themselves? Why shouldn’t customers buy AMD if for the same $ they get something better, more productive or cheaper?
AMD do seem to have (or make it) good luck acquiring and integrating technologies from other companies, so I guess they will be successful at this too.
It’s all the invisible hand of Adam Smith (famous Scottish writer on economics).
What I meant to say was, the foundry business at Intel will be strange to watch because it’s not just an enabler for their own CPU and GPU products. Will the company struggle with focus and resources between the two? I can’t tell. Will they have much luck fabbing for others in the CPU and GPU spaces? I can’t tell. Will they be capable of it from a technology standpoint? Signs point to yes. But the strategy is quite different from what they’re used to.
As you say, though, there’s so much demand that anyone who can fab something close to leading edge parts will probably find customers.
Putting that bias behind me and going NVDA along with or instead of AMD would have had me retired by now. Seriously.
Get some NVDA.
A few things on this:
- AMD is playing in a much broader market than NVDA – the CPU market, and various other data center components… Intel is second best to AMD on most of that, mostly able to crank out not-quite-competitive server parts and not-really-competitive GPUs. Intel gets what are called “Supply wins” vs. AMD “design wins” – a ‘design win’ is when your part is chosen because it’s best for the job, a “supply win” being when a vendor adopts your component for their product because they are sure they can get a virtually unlimited supply at a workable price. AMD sells all they can make, handily, in CPUs for the data center. And even in laptops/mobile Intel’s latest offerings are underwhelming (see elsewhere in the forum for the reaction to Meteor Lake) and AMD can reasonably expect to take a lot more laptop share than they’ve had so far, if they feel like it. The bigger threat to AMD in servers is I think the hyperscalers like AWS building their own CPUs…
AMD being “second best” in AI is like being the #2 runner in the New York Marathon. This stuff is HARD. AMD is gaining share in a growing market, they’ll make good money at it.
I’m not sure what the foundation of your “AMD won’t sustain itself” belief is. They’ve been awfully consistent on execution for the last several years – a very different company vs. 10 years ago, almost unrecognizeable. I am a bit concerned that the Zen 5/6 CPU designs apparently start over on a clean sheet of paper and there’s always an outside chance they will stink, but I’m not worried about much else.
So I’d say if you have Intel and NVDA, ditch the Intel and get the AMD with the money. It’s defensible to not own AMD, and to focus on NVDA, if that suits you.
I am not putting AMD down caromero, only comparing them to NVDA. AMD has done well, Nobody can say they haven’t but NVDA has done better. Everyone thinks that NVDA is only a one horse company but they have many irons in the fire. Their gaming gpu’s grew 81 percent last quarter YoY and 15 percent sequentially. The Data center that everyone is excited about grew 279 percent YoY and 41 percent sequentially. That is why everyone is excited. Professional visual grew 108 percent YoY off of a small base and Auto well it only grew 4 percent YoY. But the real kicker is CUDA, it is a software program that allows customers to run and process gpu applications. Then you have NVDA infiniband which allows high transfer rates.
Nvidia closed today at 571 after hovering around 450 most of last year. Up 14% since Jan 1. And doesn’t report earnings until end of next month.
One of the few stocks actually making earnings on AI. And a real expert at exploiting its technologies to the fullest.
No one knows if it can continue to go up like a rocket or how high but so far its a real winner. CAPs tells me in in at $15.55 and up 3572%. Kind of like winning the lottery. Once in a lifetime. I wish I could do that all the time.
I agree with the comments on how GREAT nvda is. You will get no disagreement from me on me missing the Nvidia profit. I was thinking about the past. When I bought AMD at $10, NVDA was the model I was comparing it to as NVDA climbed from $25 to $250 over the previous 5 years. My thought then was that the way AMD was dominating the cpu market, they could possibly go from 10 to $100 over the next 5 years. AMD did! Next as I watched AMD continue to dominate in the cpu arena (over Intel) i thought they could still double up to $200. Nvidia doubled up the last 5 years from 250 to over 500 currently. AMD seems on the brink of doubling up in less than 5 years - maybe 3 years. Now my perspective has changed for AMD. The AI arena carried Nvidia to new highs. So as i look at the commanding lead that Nvidia has, I agree that they are the major force in AI. I think that Nvidia will hit $1000 in the near future. That being said, what about AMD? I think AMD can capture some of this market all while continuing to dominate in the cpu arena. So far that is the case. Intel has not been able to figure out how to catch up to AMD superiority in pretty much the whole cpu arena. Now that arena is changing with the ARM devices, but it seems for the next few years that X86 architecture will still be important which continues to be the strong suit for AMD. The newest laptop cpu’s from Intel have disappointed I am reading which will be another extremely profitable area that AMD will start to see gains is my hope/thought. The supercomputer/server/workstation/desktop arena reveals that the fastest most energy efficient cpu’s are AMD. Intel still has the highest output of cpu’s albeit not as good as AMD. AMD gpu market lags Nvidia but they are seeing robust sales in this arena and there is the chance that they will continue to grow reputation and market share in this arena. The Xilinx acquisition has helped the bottom line. AMD has entered new markets that are expanding their business - automotive chips and cell phones. AMD has acquired software companies that also helped them by giving them expertise in these areas. One side note and I don’t know if this will matter, but I keep seeing news that AMD has gone with open source software for their AI chips and that because of that the major users of AI like that since they don’t have to buy the software while Nvidia does own their software that goes with their AI gpu’s and may be an additional cost. However, that Nvidia software is also screaming fast as Paul notes in the above post and so far open source is not competitive.
Regarding AI what will AMD be able to grow their small portion into. Lisa says 2 billion over the next 12 months. If this is going to be a 300 or 400 billion dollar market, can AMD get ten percent? 20 percent? If AMD grows their earnings by 10 or 20 billion, what will that do for AMD bottom line and stock price. So from the financial perspective, I can’t even guess what this AMD stock price can be over the next 2 years but I am thinking that it could be in the 250 to 300 range if AMD exceeds expectations and gets into the 3 to 6 billion dollar range. So for me $10 to $250 is pretty huge regarding the AMD share price. Paul, for me AMD has been like winning the lottery as NVDA has been for you.
I wish Bob Eachus would comment more on AMD. He was always a good source of information but seems to have disappeared. The technical expertise has always been superior on the AMD forums with all the individuals involved in the industry. Oh, I love the Saul forum that some of you post on for the financial side of your stocks, and now have been tracking those stocks as investment ideas. Thanks for the interesting conversation. I am watching NVDA for entry points…doc
The NVDA/AMD/whoever AI race is really interesting.
When I worked for NVIDIA (this was probably 14 years ago) Jen Hsun said in one of the all-hands something to the effect of “We have a GPU, they have a GPU and they’re both pretty good. The difference between us is the drivers. Ours are better and we work very hard to stay that way.” He also talked about how NVIDIA was becoming a software company. All this was long before AI exploded but also long after NVIDIA had released CUDA.
When I look at the fundamental operations that AI training and inference need it mostly boils down to multiplying matrices and vectors. Doing this at scale in hardware is not that hard. (Ok, it’s incredibly hard but there are several companies with more than enough expertise to build chips that can do it at scale.) Where things get difficult is that it’s not enough to ship the hardware - you need to provide a software stack that plugs into all the standard tools that people working on building models and consuming models use. There was a post recently on this forum that highlighted just how far AMD’s software stack is from allowing an easy transition to their hardware. It sounded grueling.
The silver lining here though is that NVIDIA is charging very high prices for their hardware and still cannot supply enough to meet demand. This is an opening for a smaller player to come in and land some customers. The key to doing that will be focusing on the software. But I think it may be possible to convince executives that they should inject a large amount of money into this type of software support to enable playing in this market. Will AMD do it? I don’t know. Will Intel do it? I don’t know. Will it come from a smaller competitor or a startup? I suspect not - the cost of entry may be just too high. But Samsung may try. Microsoft certainly has the resources and they could buy a startup for the HW if they don’t have the expertise. Apple certainly could do it but I suspect if they had something good they’d keep it in house. There’s Google and Amazon.
So I don’t think the battle is won for NVIDIA, and I think it may make sense for AMD to challenge them. Question is whether they’d view making this challenge as a strategic bet or a distraction? Former can be costly but hugely successful, latter will result in maybe some temporary sales which will disappear as soon as NVIDIA can satisfy the demand.
Full disclosure: I’m holding what is now a lot of NVDA and some AMD.
Thanks for the excellent commentary on this. I had not realized how software driven this new market is but it is sinking in…doc
For how software dependent all this is read up on CUDA vs ROCm… AMD is progressively closing the gap but it is a really large gap. Adoption is key. With CUDA a lot of software ties itself to Nvidia but porting layers are coming along … If customers implement on something higher level, which is common enough, they can be insulated somewhat.
But this really is like Intel vs AMD in that the leader has 90+ percent share and AMD can double their share and eat very well indeed without hurting Nvidia much.
I do think that AMD is going to try to challenge NVDA and there are software competitors to CUDA. I do not have any clue if they are as good or not ( OpenCL, OpenGL, TensorFlow, PyTorch, and scikit-learn), Maybe someone who uses the product could tell us, but as long as NVDA can sell everything they make, and not have to drop the price, it is hard to say anyone is competing with them. I suspect if they start slowing down and large amounts of inventory end up on their balance sheet then maybe we could start looking for a competitor. Right now AMD has around 4.5 billion in Inventory up a little from 3.7 billion and NVDA has 4.7 billion down a little from 5.3 billion so not much difference.
Anyway thanks for the discussion, I was hoping I was missing something with AMD because a lot of people own NVDA and AMD. AMD does have a wider footprint and they expect the desktop computer to start an upgrade cycle so maybe that will push them higher. Something to think about. But just for illustration, if AMD goes up 1 dollar and NVDA goes up 2 dollar why own Both? It doesn’t seem you are spreading out the risk since they both are chip companies. Here are a couple of financial sheets that I made up hopefully it will help.
Now looking at both of those you can see exactly how much better NVDA is doing. Look at the margins. NVDA has GM of 74 percent while AMD has margins of 50 percent. NVDA’s operating margins are larger than AMD’s GM and that is why I am saying just hold NVDA.
You are spreading out the risk in the sense that they are two companies battling for share in the same market. A loss for one is often a gain for the other. Obviously if something happens to the whole sector then you’re not protected at all.
I appreciate all the work you’ve put in on this and that you shared with the group. It sounds like you’ve accepted that there’s a case for owning AMD but a stronger case for owning Nvidia instead. Not at all an irrational conclusion. The time to buy AMD was when it was in the single digits and the first of the zen processors were starting to show up in benchmarks and things. If you bought then and held, for maybe 7 years, you’re sitting on life-changing money. At this point, neither of these stocks will deliver the kind of growth that AMD has since beginning under Lisa Su. Nvidia had the advantage that it wasn’t competing directly against Intel or anyone comparable. Greenfield opportunity.
One more thought on why people here are inclined to stand by AMD : Intel looked unbeatable until suddenly they weren’t. I don’t think Nvidia will make the kind of colossal systemic errors that pretty well tanked Intel for years but we have seen it happen. The business has been utterly transformed because AMD made all the right moves at the same moment that Intel made all the wrong ones.