On this episode of The Six Five Webcast, hosts Patrick Moorhead and Daniel Newman discuss the tech news stories that made headlines this week. The handpicked topics for this week are:
- Cloudera Evolve
- Lattice Earnings
- Qualcomm Earnings
- AMD Q3 Earnings
- AWS ML Capacity Blocks
- Apple Spooky Event and M3
For a deeper dive into each topic, please click on the links above. Be sure to subscribe to The Six Five Webcast so you never miss an episode.
Watch the episode here:
Listen to the episode here:
Daniel Newman: Hey, everyone. Welcome back to another episode of The Six Five Podcast. It’s Friday. It’s not the normal time. Just a few weeks ago, Patrick and I were bragging about how we’d do anything to make sure our podcast gets delivered at the normal time. Then it only took us one week to blow it, but we promise, we promise it’s your fault, not… wait, it’s our fault, not yours. Anyways, if you see the skyline behind me, that is not fake. That is a real skyline. I’m here in New York City, and both you and I have made some serious accommodations to make sure that we could be here this week because there’s a lot of tech to talk about, and we know why you’re here. Pat, how you doing, buddy?
Patrick Moorhead: Doing great. Great to be back. I think I got in my bed at 1:00 AM last night and decided to take an early flight, late flight last night, early this morning versus getting out here, I think I was scheduled at 6:00 AM. You can blame the scheduling on me, but I was ready to roll at 9:00 AM.
Daniel Newman: You were good, man. This was all Daniel’s fault. By the way, I just want to point out that that’s the second time today I’ve referred to myself in the third person. I think if I do it one more time, there could be some sort of condition that would need to be diagnosed, not by anyone on this call, and hopefully, not by anyone listening. But, in all serious, it’s been a very busy week in tech, and we got a great show for you today. If you haven’t listened to The Six Five before, I’m going to ask you what’s wrong and why, and then I’m going to remind and tell you that this show is all about analysis. We cover some news, but only as little as we have to. There’s plenty of places to get the news. We are here to dive in between the lines and talk about what’s happening.
It’s been a big week for earnings. It’s been a big week for New Silicon. It’s been a big week for some events. Cloudera evolved where The Six Five was on the road rocking out, and then there’s some interesting news from AWS that caught your attention, Mr. Patrick, and you’re going to talk about and I’m going to then opine. We’re going to hit a lot of things. We hit Cloudera, we hit Lattice Semi. We’re going to hit Qualcomm, we’re going to hit AMD, AWS and Apple, and we are excited to do that with all of you. Now, quick reminder, this show is for information and entertainment purposes only. While we will be talking about publicly-traded companies, please do not take anything we say as investment advice. All right, take a breath. We did it. We’re through the introductions. Pat, you got back from New York, but not without an update from Cloudera. Tell us what was going on at Cloudera.
Patrick Moorhead: I don’t even know what that question means or what that introduction means. You and I were both in New York City for more insights and strategy, Futurum Group, and The Six Five. So we took in all the analysts’ content, but we were also there… we had five interviews with multiple executives, including the CEO, a few partners, but let me break this down here. Cloudera is a data management company, and there’s a lot of companies that say that that’s what they do. For instance, you have ransomware protection, you have data protection. You have security companies calling themselves data management, but Cloudera is a little bit unique, and it tracks all the way from the inflow of that data, cleaning up that data, many times storing that data, let’s say in a data lakehouse that’s based on Iceberg, and then getting it ready for analytics, machine learning, deploying inference to the edge and then being able to track all that data, who should have access to it, the lineage of it, how it changed, who has it.
So essentially, if a regulator comes in like they do, particularly in finance and healthcare that says, “Hey, I want to every bit of data that you have that’s in the entire enterprise, who has access to it? How are you protecting it when it changes?” The key here for this event was this was their big AI, generative AI play. I think people who understand this understand that garbage in, garbage out, you’ve got to have data, the right data to put in there, but things really changed with generative AI. While most of generative AI is domain specific, really, it’s going to be a future that is going to be commingling ERP data, CX data, finance data, confidential information, public information, and that exasperates the data challenge.
So there were really one big announcement across three players, and this was an official partner network, and Cloudera introduced AWS, Pinecone and Nvidia. A lot of the conversations on why they chose them is industry leading; industry leading GPU, industry leading IAS provider. At least by every metric I’ve seen AWS is the AI, IaaS leader in terms of the amount of revenue that they generate, Pinecone, when it comes to vector database and Nvidia being Nvidia, training and inference. But first and foremost, I hope the takeaway was, and this is not a new takeaway for us, was that if you have a hardcore AI or generative AI, you need a data management platform that spans all of that.
I’ve been very vocal that it’s been a big disappointment from a lot of on-prem plays that they don’t articulate what the data management platform is, is they just ignore it. While we would expect Cloudera to lead with that ’cause they’re a data management company, that doesn’t not make it true. It is true. Oh, by the way, the public cloud folks all have an end-to-end data management platform. Their biggest challenge is, “How do I get the data on-prem?” The number you and I throw around 75 to 90% of enterprise data being on-prem, how do you activate that in the public cloud? So anyways, I’ll leave it at that. It was a big event. I think Cloudera has a big opportunity ahead of them. I think they’ve got a big marketing challenge ahead of them as well. They’re a PE-owned firm, and they’re going against folks who can at the same revenue level, can spend 10X more on the marketing than they can afford.
Daniel Newman: Yeah, I think you hit a lot of the really important points. First of all, it’s a new look Cloudera, starts with the management team. Charles Sansbury is CEO, he’s bringing in new leaders. If you go to year to year, it’s a very different look, feel. The company is making some commitments to get away from its deep engineering geeky roots that I think have plagued it a bit in terms of adoption. At the same time, the company still benefits of being a majority stakeholder of data management for five plus of the largest companies in every major industry which are using Cloudera. Now, I’ll say it here, I’ll say it there. They need to show that A, they can create growth with those customers. Meaning it’s not just legacy Hadoop, it’s these customers are going to use and expand their current spend, so it’s land and expand, and then it’s new customer acquisition.
If Cloudera really is up with the times, which we believe it is, I believe it is, it needs to be able to show its muscle by winning new big customers that say this is the platform we want to trust our future to. We want to build our company’s data future to be able to solve AI, generative AI, analytical AI problems. So I felt that was on display. Some more customer cases, more partner focused, a little less geek, but at the same time that developer ecosystem has to be served, so it’s a nuanced challenge.
My early inclination is to say Charles seems prepared to play the role in that leadership space. I think in our conversations with him, he seems to be on the right track, saying the right things. Also, it was good to meet Mary Wells, their new CMO. She brings a wealth of experience and seems to buy into a lot of what you and I have been saying the company needs, which is those case studies. It’s those reflective large user wins and the ability to articulate those from a value and solution standpoint, not just from the technology. So good event, good to be there. Good to see you there, buddy. Sad you’re not here with me right now today, sitting side by side with our sunglasses on in the Hawaiian sun. Oh wait, I’m in New York, forget it, but–
Patrick Moorhead: By the way, for what it’s worth, I didn’t like that sunglasses look. We looked awkward, and I can’t figure out exactly why.
Daniel Newman: I thought you looked great. You looked like the new thin 210 Pat that I’m so proud of.
Patrick Moorhead: Aw, bestie.
Daniel Newman: Now for me. Yeah, I looked like I’d been on the road eating, what would be the word, unlimited… If Old Country Buffet was still open, I look like I’d been visiting.
Patrick Moorhead: Yeah.
Daniel Newman: Anyway, let’s move on. Next topic, we’re going to hit a few earnings. Let’s start with Lattice Semi. Pat, Lattice has been a multi quarter darling of this chip run and chip glut and then this chip shortage. Frankly, it was a company that I think we credited Jim Anderson and his leadership team with running extremely well throughout tumultuous times, getting into the right markets, and really creating a strong lead in the low end FPGI space. The company had another good quarter, record revenue. Now, no, it was slower growth, slower sequential, good margin performance, solid year-on-year earnings per share. But like so many companies, sometimes that growth, beat, raise musical chairs game does stop. This was the quarter where that music did stop for the first time, and I think it was like a dozen quarters. The guide was conservatively and down for the company, and this is really associated with the industrials. Industrials, say that again.
The company has rallied with the strength of industrial, but we actually saw in Qualcomm’s numbers, which you’ll talk about, Silicon Labs numbers, which were both IoT industrial and…. that market just got clobbered. Now, frankly, not the case for Lattice, which had solid performance, but it did slow. That slowing is what’s causing the conservative guidance, and that conservative guidance caused a pretty negative reaction on Wall Street. But coming out of these earnings, I just got to say, Pat, I think the company is really well positioned. I think they’ve been conservative. They’ve guided correctly. They’ve been building the right products. They’ve been getting into areas like AI, FPGAs for things like computer vision and other critical designs for security in laptops and automotive, in servers. The company’s also expanding into that middle class. It feels like a middle class of FPGAs that doesn’t really exist in a significant way.
You have that really high end, the Xilinx part of the market, the PSG part of Intel that’s spinning off. But on the lower to middle end, Lattice seems really well positioned to gain market share. It’s really just ramping. It’s Avant product. So while, of course, we always want the guide up, you know the drill, Pat, beat, beat, raise, beat, beat, raise, the music can’t go forever. But I actually think overall, it was a solid quarter amidst some pretty significant headwinds. If you saw that jobs report that came out today, I think we’re going to find out at some point that a lot of the data we’ve been getting is horse poo-poo for a while now. The feeling in the market isn’t nearly as good as what I’m actually seeing in the market. Starting to see some of these slowdowns, some slightly weaker guides is probably a good thing. Also, starting to see turns in some of the things that have been incredibly weak over the last few quarters, years is also a good sign because we know the economy is cyclical.
Patrick Moorhead: Dan, you left me a little air, not a lot, but-
Daniel Newman: Talk about anything you want.
Patrick Moorhead: No, listen, none of this is self-inflicted other than if you look at 14 quarters of revenue, gross margin, EPS and operating profit improvements as self-inflicted, the markets that are down they were down in and there’s really… they’ve done a decent job making up increasing market share and B, getting into brand new markets that’s so far… don’t yawn on me, buddy. They’ve been able to overcome at least eat the yawn. Geez. It’s like, “Oh, it’s so inter….
Yo, listen, 5G telecom down, not a shock, all of IoT, first it was the consumer IoT and then it was the industrial IoT. Auto up with many automakers except for Tesla as we saw with ON Semiconductor. So great lineup. They’re increasing their TAM with Avant by at least 50%. So it’s not like some big conundrum of where they’re going to get their growth from. Markets go up, they’re going to go up. They’re going to continue to gain market share in areas that their competitors have left uncovered and even areas where they’re converting a low-end processor or microcontroller to an FPGA. So all in all, I’m not panicked and nobody else should be. I don’t trade in the stock. I don’t trade in any of our paid clients or anybody I’ve signed an NDA with, so…
Daniel Newman: Do you want me to do the disclaimer again just for everybody?
Patrick Moorhead: Sure, yeah. Let her rip, buddy.
Daniel Newman: This show is for information and entertainment purposes only. While we will be talking about publicly-traded companies, please do not take anything we say as investment advice. While we do that, let’s talk about Patrick Moorhead on CNBC kicking A and taking N. You went on with the always impeccably-dressed John Forte and Morgan Brennan. You talked a little bit about Qualcomm, but let’s talk about it here.
Patrick Moorhead: Yeah, so I actually was on CNBC twice in a week talking Qualcomm, and I really do appreciate being on Morgan and John’s show. So Qualcomm absolutely blew away expectations. It was beat, beat, raise, which was super fun. Surprisingly, they talked about smartphones, which was the big driver, and also automotive, which really hasn’t made the bar aside from their $30 billion backlog. So really impressive. While the company Cristiano and Akash, and I did appreciate talking to them about earnings, didn’t necessarily want to call it the end of the bottom, but I believe that this is the end of the bottom here. They need to be very conservative because they don’t control the China economy. But from all my channel checks and all the data that I’ve seen, the inventory, particularly Android inventory has gone down. Oh, IoT was soft just like everybody else when we talked about with the context of Lattice, but great tailwinds in handsets.
It’s interesting that we saw Xiaomi and a couple other of the big Chinese handset makers launch their handsets the same week that Qualcomm came out with Snapdragon 8 Gen 3. That’s actually a pull-in on the dates they normally do that ’cause typically, you’ll see Qualcomm announced in the fourth quarter and then go live with units in January. I see that as an acceleration, which I think will benefit the company going forward. I have to talk a little bit about their new PC platform, the Snapdragon X Elite. We’ll talk about it a little bit when we get to the Apple event, but they ran up some incredible numbers versus Apple, Intel and AMD. Now, I did not run the benchmarks and neither did Futurum Labs, but at least the methodology looked sound enough for me to at least say it’s going to be a very competitive part. We’ll talk more about that when it comes to the Apple event and M3.
Daniel Newman: Yeah, so you did hit it on the head, Pat. Qualcomm was robust, and again, this is a comparative robustness that we have to talk about. I think it’s only right. Year-on-year, it’s still not where the company wants to be, but we have to be very clear about the headwinds. You’re talking about a company that had over 60% of revenue relationship with China and in an economy in China that’s become incredibly uncertain and to some extent, unstable. Mid double-digit growth in China handset, which means, again, while the diversification strategy, I think, has worked very well, especially automotive, very impressive transformation there under Nakul Duggal’s leadership. But in handsets, it’s been pretty woeful and especially in China. The concern has been palpable and the stock has seen a response. But again, this is, Pat, something I think you said about Lattice. This wasn’t much self-inflected.
When you talk about the innovation and the design and what we saw at Snapdragon Summit, the company continues to out-innovate, out-design, deliver better specs, better SOCs, and better performance to all of the handset makers. This is just a byproduct of less people buying handsets. Even in Apple’s numbers, you saw the Device softness. It wasn’t like Apple’s getting all the market. Nobody is buying smartphones. Now, nobody as in only like a 100 billion this quarter, but I know, dollars not units. The point, though, is the growth is coming in the right places, the turn, the bottoming out of inventory. We’ll talk a bit more about that when we hit AMD here in a moment. But we’re seeing a transformation of where the bottom is. So we saw enterprise stand longer. It’s been more robust because of AI.
We saw industrial devices stick around a little bit longer because companies were implementing technologies to measure analytics. Handsets, consumer stuff fell off the cliff at the end of the P word, and it’s really only slowly coming back. But the slow revival for many people out there that have questioned should be a good indicator, not just Qualcomm, every company in the handset space, all the OEMs. You look at companies like Dell, HP, Lenovo that have had long, slow periods from their device parts of their businesses that should be turning a corner because silicon is often the leading indicator of when those cycles are going to begin. There’s a lot of indication here that things are turning. So no one is saying so much as a bottom, Pat, but you and I have talked to a few CEOs this week, including Cristiano Amon and the positive tenor in their voice is encouraging.
Patrick Moorhead: Yeah. Isn’t it funny what people can tell you without telling you?
Daniel Newman: That’s the art, right?
Patrick Moorhead: Yeah.
Daniel Newman: Actually, I think that’s, we’re professional tell you without tell youer. When you have so much information rattling around in your brain and you’re never 100% sure if it was NDAs, so you just say nothing.
Patrick Moorhead: Yeah.
Daniel Newman: I got an NDA filter. I need a Gen AI in my eyeball. It’s like, “Nope, that was NDA. Don’t say it.” Anyways, good job, Qualcomm. Let’s rip on one more earnings, Pat. Let’s talk about AMD. You and I actually ran into and had a little time to sit down with CEO Lisa Su when we were at the Dell Tech Analyst Summit. The reason we’re not talking about the Dell Tech Analyst Summit, by the way, is ’cause it was all NDA.
Patrick Moorhead: Right.
Daniel Newman: So we’re not going to say anything, but let’s just say we did have the chance to spend time with Michael, with Lisa, with a bunch of others, Jeff Clark and leaders from the Dell team and encouraging to listen to them talk about AI. Let’s talk about AMD. Okay? Long and short, it was a beat on revenue, margin and EPS looked good too, but people didn’t love it. Why didn’t people love it? Because the only thing people care about when you beat is your guide. So when you beat and you guide up, they’re happy when you beat and you guide down, they’re not. It really doesn’t matter how much you beat by because it’s the, “What have you done for me lately?” Syndrome. So everybody’s wondering, now, this is a really weird kind of a tale of two guidances. I think both of you just simultaneously looked at Lisa and congratulated her on the really encouraging guide around AI, the GPU data, GPU number. What was it? It was like 2 billion on the guide of GPU for their MI series?
They were telling a very compelling story that moves revenue off potentially of Nvidia at the very least, and enables customers that want an alternative to work at the higher layers of abstraction and frameworks with AMD to be able to get workloads on their silicons pretty seamlessly. At the same time, because gaming’s pretty weak, parts of PCs are still only just coming back, there was a very near miss on guide and everybody went ballistic about it. But Pat, I want to talk about a couple of things besides just the beat raise. The company’s operating very well. So if you didn’t see their margin, I think it’s up 5%, what a tremendous growth on a year-on-year basis. An operating margin saw a significant 5% increase on year-over-year. These are good things. When companies are in tougher periods of growth. I always look to see if they are operating well?
So we’ve seen this throughout the past few quarters when companies are trying to find profit, it’s, can they also manage the budget well? Can they manage operations well? Can they manage hiring well? Can they manage production well? Can they manage their supply chain well? So there’s a lot of indication inside the business that it’s not only running for the top end, but it’s running to become a much more efficient bottom line production machine. This is a company to be excited about, Pat. This is a company that has a good opportunity to be the number two AI company. It’s made some very important acquisitions. The Xilinx acquisition, for instance, was very important. It’s also making some pretty important moves around AI. It’s making some tuck-in acquisitions, send out some very interesting partnerships. Pat, this is a company to be encouraged by. I know you’ve got a little bias here ’cause you were, of course, the strategy chief corporate development guy. I saw your street cred.
Patrick Moorhead: Don’t even go there, dude.
Daniel Newman: I saw your street cred. It popped up.
Patrick Moorhead: I don’t have any bias, dude. I’ve been gone from that place for 12 years.
Daniel Newman: Sorry.
Patrick Moorhead: I got fired from there.
Daniel Newman: Sorry, buddy. Yeah, let me be clear about that. You’re very reputed for covering this company; not a bias of like or dislike, just you’re very well known because of your experience.
Patrick Moorhead: I know I heard the word bias, so I’m reacting to that.
Daniel Newman: Yeah, no, that’s fair. I need to be called out. I should be called out for that. I mean you’re known for your role there. It was a street cred joke. It just didn’t land. Let me do that again. I saw your street cred the other day with Forte where it pointed out that you were the AMD guy, okay, back in the day.
Patrick Moorhead: Yeah.
Daniel Newman: We blow that out. Let’s blow that out. Anyway, point is that solid, solid, solid, solid guide, eh, AI matters. That’s what it is. AI matters, can they land the AI thing? That’s probably my only gripe. My biggest gripe is just like, okay, for a year now, I keep hearing when is it going to start landing?
Patrick Moorhead: Yeah. So good coverage here. I’m going to hit some of the data center highlights. So Dan, you correctly talked about the company waiting for what they could do with AI. On the call, Lisa announced that they’re going to see $2 billion of revenue in 2024 based on GPU accelerators. The MI 300 is going to be, this is per AMD, the fastest to $1 billion that any product they’ve ever done. When I think about Opteron, I think of Ryzen, I think about Epic. I think maybe I said Athlon before, but it’s shocking that this would be the fastest to billion. Quite frankly, when I talk to CSPs and even the enterprise on-prem folks, AMD is the clear number two in AI accelerators. It comes down to a GPU design versus an ASIC, and I love ASICs. They’re harder to program too. They’re more efficient than GPUs, and there are little blocks of ASICs that are inside of what people call GPUs, but it’s just easier to program.
Intel does have a data center, GPU, and I’ll admit, I’m trying to figure out why people don’t view it as a second or even a third. It appears to me that the market is getting more excited about Habana 2 than anything else. My instinct, no pun there, tells me that Intel GPU is architected more for high-performance computing GPU than it is for AI training or AI inference ’cause quite frankly, we just don’t hear a lot about it. Indicator that the market is back when you look at the PC market is back, when you look at what Intel pulled off, but the client segment was up 42%. Okay? So we saw Intel, we saw AMD, that inventory, I can confidently say has been flushed out of the ecosystem and look like it’s doing well. Not going to comment on gaming or embedded just because they’re not what people are focused on. The gaming client revenue, its largest client, I just said I wouldn’t cover it, but I am, but yeah, the revenue was down.
That’s really on the back sell in of the gaming consoles. I was cited by AMD and a couple of the other things as well. Radeon GPU is actually up, so I’ll skip Embedded. Well, I won’t skip Embedded. They were down. Intel was down and so was Lattice, so no surprise. Industrial markets impacted, data center stuff is questionable. I’ll move on. But AMD is in the game on AI PCs as well. We haven’t heard a lot of details out of it, but they claim to have 50 notebook designs with rise in AI in the market today. They announced a really cool CPU architecture that they announced with notebooks that, I don’t want to call it big little because that’s arm and I don’t want to call it, but even P Core and E Core that Intel is doing, but there’s a rise in 4 Core and a rise in 4C. The C is the small core and the rise is the big core. I’m hearing that it’s easier to program than P and E Core, but I don’t know that myself. I haven’t done the analysis and neither has Futurum Labs.
Daniel Newman: Big Pat, little Dan.
Patrick Moorhead: There we go.
Daniel Newman: Big Pat, little Dan. So that was actually some pretty good insight, and that’s something that I think our performance labs should probably tear into. We are tearing deeply into the AI PC. You can keep an eye out for some of that. That will hit early in 2024. That’s going to be a trend line we are going to be talking about a lot. Pat, AWS dropped some interesting news this week. Talk about what they’re doing with their machine learning Capacity Blocks.
Patrick Moorhead: Yeah, it’s interesting in all this world of generative AI and all this stuff, it’s like gosh-
Daniel Newman: Are you still ML?
Patrick Moorhead: Are we still doing ML? Yes, we absolutely are doing ML and quite frankly, for narrower data sets than generative AI, which can be up to 100 petabytes at this point just for the training model, it’s more efficient and less expensive. So I saw that, Dan. I saw that.
Daniel Newman: I bit it.
Patrick Moorhead: Okay. No, this is good. But one of the challenges for let’s say smaller businesses and even startups is, how do I come in there and reserve enough GPU to do what I need to do? It’s not just one GPU. When you’re doing machine learning training, let’s say you’re trying to train a vision model or something like that, you need hundreds of GPUs that are interconnected in a logical way that have a singular memory plane to be able to do all that work on. So what Amazon did is they brought out what’s called Capacity Blocks for ML for machine learning workloads. What that does is you can actually schedule like a hotel, which says, “Hey, on January 5th, I need this much capacity for my GPUs maybe for this long.”
What they do is they reserve that, and these are EC2 Ultra Clusters, which means that they’re of the highest performance optimized for ML. Then they’re connected through what’s called EFA, which is Elastic Fabric Adapter, because we all know that it’s not just about what you can do on that rack, but what you can do across multiple racks. EFA gives you what they’re calling petabit-scale, non-blocking network, which essentially is to get your workload done a lot quicker, again, to have that plainer memory. So that’s it, baby, MSL Capacity Blocks. Look at this space. I see no reason why they also wouldn’t do 1000-node systems for foundational models, but that’s just a prediction.
Daniel Newman: Read my lips, this will get bigger. Pat, why I think this is also important is, I don’t know about you, but I’m really getting tired of having to always say, “I know everyone thinks AWS is behind on generative AI.” You know what I’m saying? I feel like if I don’t start…
Patrick Moorhead: I know it came up in even our interviews yesterday.
Daniel Newman: It came up, and I’m going cross-eyed. It’s like, look, AWS has more workloads on more computes across more geographies than I think pretty much all the hyperscalers combined. Now, again, that’s funny math, so don’t hold me to it. But the reason I’m pointing that out is computes and data and workloads are the impetus of AI. So AWS had a different strategy, and it wasn’t based upon necessarily completely locking it up with a single large language model in a closed architecture, which some others did early on, and we were able to get good out-the-gate marketing and story narrative leads. But AWS has taken that we’re going to be the open source. We’re going to be open in our approach, and we’ll offer some FMs, foundational models. We’ll offer Titan. We’ll offer Bedrock. We’ll offer API connectors. You can run multiple models same time, all that kind of stuff.
So Pat, I just chalked this up to AWS being AWS, offering lots of services, looking at how did you take all of its capacity, add value, and make it easier for companies to do the things that are going to need to be done. As we know, there’s going to be more and more smaller models, middle-sized models, foundational models, and the large language models are increasingly commoditized. We’re seeing a world where hugging faces GitHub for AI, and that’s where this is going. There’s competition for that, but I think it’s a good use of resources. So I don’t have a lot more to say about it. It’s early, so I think we can save ourselves a bit of time, Pat, and talk about the Spooky Scary Event that took place.
Patrick Moorhead: Spooky Scary Event.
Daniel Newman: Scared of the M3, 18 TOPS being the next-gen IPC. All right, well-
Patrick Moorhead: I don’t think we can take it any farther.
Daniel Newman: No, I got nothing.
Patrick Moorhead: No. Listen, Apple scheduled a 30-minute event during an NFL game and during the World Series the day before Halloween. I guess it could have been even worse. It could have been on Halloween, which distinctly indicated to me that this was a complete and total rush job to respond to a couple of things. First of all, was Qualcomm coming out and doing the smack down and showing that these were measured by Qualcomm and not more insight than strategy or Futurum Labs, but they were industry standard benchmarks that they let us watch them perform.
They showed this incredible performance per watt, not just with the 45 TOPS, but also with this incredible CPU that they put in here that, Qualcomm didn’t have the numbers, but showed that it outperformed… sorry, didn’t have M3 numbers, but showed that it outperformed pretty much all of the M2 line. Then Apple’s rush Spooky Event was also a reaction to what Intel is doing where they’re talking about the AI PC where CEO Pat Gelsinger was talking about on earnings calls. They made an announcement about 100 ISV, I saw that, Dan, that would support it and-
Daniel Newman: Okay. Okay, everybody.
Patrick Moorhead: Yeah, they would support that, and then at the Intel Innovation event, Intel talked about it as well. So all these things going on, you’ve got AMD talking about their AI PC. So Apple had to feel some pressure. By the way, folks, according to some industry analysts that are not mine or Dan, Intel units, sorry, Apple units went down 24%. I saw that as not that people don’t like Macs, but they stuffed the channel the quarters before because these analyst reports are about sell in. They are stuff that was sold into the channel, not consumption and sales out. What we saw yesterday was that Mac revenue went down 34%. So you can imagine all the pressures of this, but let me cut to the chase. M3 looks like the same micro architecture as M2 and M1. It is on a potentially improved process, which is the, quote, unquote “3nm TSMC process.” We saw some challenges with the iPhone 15 when it came in with their quote, unquote, “3nm TSMC process-based processor,” that led me to believe that they were having issues because they’re overheating.
Just because they could fix it with a software tweak, it doesn’t mean it wasn’t a chip issue ’cause how often does Apple screw this up? You do thermals and you do software around the certain guardrails of your chips. So hence we have M3 where I expected just a desktop, but we got an iMac plus a MacBook Pro. We did not get errors. When I step back, and I look at what’s going to happen in the industry in the middle of the year, I believe I’ve seen numbers, again, I didn’t run these numbers myself, that Qualcomm, Snapdragon X Elite will outperform at least the M3 Pro version on single-threaded and multi-threaded CPU benchmarks, didn’t run it, just speculating. It appears that the Snapdragon X Elite will have 2 1/2 times the NPU performance as the Apple event. The reason I’m putting so many disclaimers here, folks, is not just that I didn’t run any of these numbers myself, but that there are no independent third-party reviews of any M3-based product.
So net-net, I still think that Qualcomm looks really good mid-year. I think that this is not some M3 that comes in and wipes everybody up again like we saw with M1, and it makes sense, right, Dan? You pointed out the AMD thing that I used to work there and a couple of other things, but here’s what I know. Here’s what I learned over many years. You can’t stretch a microarchitecture too far, three variations, and then you can make it smaller. You can attach them together like Intel has intelligently done. You can go to new packaging, and the competition still has yet to do integrated memory like Apple does that gives it a little bit of an advantage there. But it is getting long in the tooth, okay? I would expect that this Rev4 M4 would have to be a brand new micro architecture. TSMC is slowing down. It’s hard to get the benefits of doing transistor-level architectural shifts. You could throw a bunch of more cash on there, but cash means it’s a bigger chip. Anyways, I’ve gotten so geeky, I just need to stop.
Daniel Newman: You covered my topic well. Thank you.
Patrick Moorhead: Oh, my gosh. Was this your topic first?
Daniel Newman: Who’s counting? But the-
Patrick Moorhead: God, I’m so sorry.
Daniel Newman: No, that’s okay. I paused for a minute and you jumped right in there. So look, I actually will just make a couple of little ads and then we’ll send everybody on their way. First of all, this seemingly was a reaction, and we can speculate, but a reaction to the impressive outputs from the week prior, just based on when it was booked, when it came out, and the fact that it was awkwardly timed. To their credit, I actually wish they would do all their events this quickly. I’ll watch it back.
Patrick Moorhead: 30 minutes.
Daniel Newman: I don’t need two hours. You do it all in 30. Let’s speed it up ’cause for me, I like to consume it in less time. I think your point about microarchitecture is very important. I think that clearly there’s some issue with TSMC and what they’re getting out of the 3nm. It just doesn’t seem generationally like it’s that big of an improvement. To your comments about the Mac numbers, it’s a little scary out there. That’s not good performance.
Patrick Moorhead: What’s scary is a 34% revenue decline.
Daniel Newman: 34%. But to your point, they held stronger when others were doing really bad, so at some point, maybe the theme of this is the music has to stop.
Patrick Moorhead: Yeah.
Daniel Newman: The music has to stop. I think at some point, it’s their time now. They had that eighth quarter of revenue decline or whatever it was, or fourth quarter, I can’t remember what it was. It’s a big difference, by the way, four and eight. But I know it’s a multi-quarter decline, and that is alarming. Having said that, by the way, unrelated to the event, Apple did see 16% growth in services. You do know that the hardware era is going to end, and that is the one part that I wonder if they can get that going. By the way, Pat, 16, I know, not that exciting. Look at you.
Patrick Moorhead: No, no, no, no, no. What do we know now is that the, I don’t know, $20 billion that Google pays Apple is part of that service revenue, if not all of it?
Daniel Newman: It’s a pretty important one, ain’t it?
Patrick Moorhead: No. Isn’t that it came out, so there was a redacted-
Daniel Newman: Yeah, oops.
Patrick Moorhead: But what came out through the Google FTC, I’m like, I’m 80% sure of this, by the way, but I tracked, where else would it show up?
Daniel Newman: Yeah, no, it’s material. It’s material for sure, and so that’s going to be an interesting part of their business to watch. Pat, one big thing too is this, the neural processor doesn’t cut it. It just doesn’t, it’s not going to cut it for what these AI PCs are trying to do longer term. Now, again, Apple might have another announcement in three months, and it might be at 50 TOPS or whatever. But at this spec, it doesn’t seem to be very compelling in that particular arena, which is, it’s going to be important. I guess we’ll see longer term. But I’m not running to a store to buy an M3 If I’m an Apple customer, and I’m not a big… I have an iMac. I’m not seeing generation-to- generation improvements that encourage me to run store to store. But let’s face it, they haven’t been doing that with their phones either. So I guess for people that care about performance, maybe that’s not really where Apple wants to play. Someone said to me the other day, “The people that buy Apple can’t spell Qualcomm.”
So the question mark is always going to be, are they going to be the best marketing company? And they probably are, probably have and probably will be. But from a technology standpoint, I’m going to yawn one last time before I end the show. All right, buddy. Then there we have it, six topics, not five minutes each. The Six Five brings it every single week. Sorry about those yawns. I gave you the sixth topic because I felt like I owed it to you after falling asleep your last two times you were talking. But in all seriousness, another great show, great week. It’s great to see you, spend some time together. I’m heading to VMware Explore in Europe for a couple of days. Then you and I’ll meet back in New York again for IBM’s research event. So stick with us, stay with us. We appreciate you. We appreciate all of you out in the community. For this week, for this episode, for Patrick Moorhead and myself, we’re out of here. See you. Bye-bye.