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:
- Qualcomm Snapdragon Summit 2024 Recap
- ServiceNow Q3FY24 Earnings
- Arm Pulls Qualcomm License
- IBM Q3FY24 Earnings + Granite 3.0 Reveal
- NVIDIA Blackwell Design Flaw Fixed
- SAP Q3FY24 Earnings
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 on your favorite streaming platform:
Disclaimer: The Six Five Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.
Transcript:
Patrick Moorhead: The Six Five is live. We are back, still breathing after a coast to coast, island to island hop. Couple red-eyes in there. Dan, it’s great to see you my friend. This is the relaxing part of the week. We’re almost at the finish line. I guess maybe the finish line are Saturdays because we just don’t know when to say no.
Daniel Newman: Yeah. There really is no weekend, but there really are no rules if you can plan your life accordingly, that’s the beauty of entrepreneurship. But yeah. Look, it’s been a heck of a week. We did do some travel. There was a 41-hour-long trip to get me from Austin to Maui. I got to go home for a few minutes in between. Broken planes, missed flights, re-bookings, new connections, an F1 race was somewhere tucked in the middle of there. Didn’t get to go to the race race, but went to the pre-race. But hey, I am back. I didn’t get a tan either. So I was in Hawaii, no sun, no tan. But other than that, I’m doing good. How are you buddy?
Patrick Moorhead: I’m okay. I got up at 3:00 AM on Monday morning, East Coast and made my way out to Maui. That was about a 20-hour trip. Was out there long enough to… Actually, I didn’t even have time to smell a rose, but I did go to some great keynotes, met with some of the Qualcomm execs out there, like you did. And then I actually saw you on the red-eye flight coming from Maui into Dallas. I saw you, were we on the same flight?
Daniel Newman: What a funny, funny coincidence. Man plans, God laughs. Works in a funny way. On the second leg we were even next to each other and people probably think we did it on purpose, but it was just crazy happenstance. Now sadly, it was like one of my best tweets of the week was when I said, “Hey, it’s you.” Instead of the long, thoughtful stuff that I write about the future of apps, AI, silicon innovation, China, because we know people are stupid. Not you, the watchers, because if you’re here you’re smart, but the higher quality of the content, sometimes I do one to lower odds and anyone’s going to read it and pay attention.
Patrick Moorhead: Exactly. So let’s dive in here. We’ve got a great show for you. We’re going to be talking about the Qualcomm Snapdragon Summit 2024, Six Five was there meeting with senior executives from Qualcomm. And we had a special guest that they chose Dan to interview over me. And not that I brought that up, but I think it’s because I’m a huge Red Bull fan. But anyways.
Daniel Newman: You’re pretty mature about these things too, and that’s what I really like about you.
Patrick Moorhead: No, and that’s the great thing about being 56 is you get mature and you don’t carry over any of those adolescent bad FOMO things. But anyways, we’re going to hit the summit. We’re going to hit ServiceNow earnings. We’re going to talk about a related topic to Qualcomm, which is Arm decided to give Qualcomm 60 days notice. We’re going to be talking about IBM earnings and they dropped Granite 3.0 models for enterprise. Nvidia, I think this is the end of the conversation about the Nvidia Blackwell design flaw. It is fixed. Is it fixed? What happened? And then we’re going to end this segment with SAP earnings. Daniel, and we’re going to do what I know you love, which is I’m going to call you-
Daniel Newman: You call your own number?
Patrick Moorhead: Call me a number.
Daniel Newman: I’ll call my own number.
Patrick Moorhead: Yeah. So Qualcomm, they’ve been on a multi-year revenue diversification plan. Gosh, for the past year plus, they’ve talked a lot about how they’ve grown in the automotive market, that gigantic 50 billion backlog. And also getting into the PC market where quite frankly they have redefined the segment. I don’t use those words loosely, but they came out with their very high performance, low battery life designs. Both AMD and Intel did respond with offerings of their own. But this Snapdragon Summit really the lead here was the new mobile core, the new Snapdragon 8 Elite, which is based on a second generation Oryon core. And Oryon 1 is inside of the PC, but they made some pretty incredible modifications. And although we didn’t do any of the testing, they did let people test on some basic synthetic benchmarks.
And for the first time in many, many years it surpasses Apple on a few key benchmarks. It meets them on single threaded benchmarks, but pretty much blows them away on anything that’s multi-threaded. The GPU and the NPU aren’t as clear, but with phones going on sale you can benchmark them in the next month. That was the big news there. By the way, it puts them in a very good position. They already dominate premium Android, but there is a potential particularly outside of Western countries where they can potentially move a few points of market share related to Apple. Also piqued everybody’s interest is would these next-generation Oryon cores and design make their way into the next generation PC? The next big announcement was around automotive, where we saw, imagine that, the new Oryon cores going into not only the cockpit solutions but the ADAS solutions. We had a great sit down with Nakul who runs that business among others.
But it’s too early to see necessarily how this does in comparison from let’s say what Nvidia is doing or what Intel is doing. But what I can say is that monster backlog they have, which is up to $50 billion, is proof positive that their strategy is working. And the way that I would characterize their strategy is meeting customers where they are. If they want an integrated solution, top to bottom hardware, software, and services, they can offer that. If they only want to take bits and pieces of the solution and leverage their own software, they can do that too. It is so funny, just remembering back even a few years ago before Qualcomm had even announced that it was doing ADAS, they invited me to do a ride in a car that was driven by Snapdragon processors.
And then here we are, I think roughly four years later, with this monster backlog that it really has three parts to it. You have the communications right? Which has been the 3G, 4G, 5G communications, Wi-Fi, Bluetooth. You have the in-cockpit, essentially everything you see and all the entertainment, services that go along with that. And then finally ADAS and full self-driving. We haven’t necessarily seen the full self-driving in a car yet, but I am sure we will as Qualcomm has put together what I believe is a scalable solution to get in that market. The other thing that we don’t necessarily think of that could come out of this group are things like robots and those companies who aren’t doing their own chips, like the folks over at Tesla. Robots, two wheel devices like motorcycles and even think industrial. So hoping to get a lot more information financially from automotive, PC, and the mobile market coming into Qualcomm’s Financial Analyst Day, which is next month.
Daniel Newman: Yeah. There was a lot to unpack. This wasn’t unpacked, but Qualcomm did share quite a bit. And it kind of all started off with a bit of a storytelling exercise by CEO, Cristiano Amon. We had him on our show and we will publish that probably in the next few days. But it really started off with him talking about what AI in the future looks like, which was a bit unique. This is a company, remember a couple of years ago it was like banging on 5G, it was banging on SOC advancements, features, images, photos, and this was an all AI show. I don’t know how much you remember, go back two or three years, and just how different the content has become. He was talking about agents, he was talking about on-device, he was talking about LLMs running locally. And that was the meat and potatoes is you can sense that the hardware industry is becoming cognizant of the narrative and the impact that AI is going to have and how they’re going to need to position themselves to stay differentiated. And so that was very evident.
And then, of course, the company did focus on how Oryon is going to become platform distributed. So whether you’re going from phone to PC to automotive, it is going to be building continuity. And this really aligns to that diversification story that the company’s been telling over the last couple of years. You and I said for a long time, too much risk in handsets, too much risk in Apple. It’s really done a good job of being able to offload those concerns. Of course, there’s some new risk related to its ongoing dispute with Arm and we’ll hit on that a little bit later. But this Oryon platform that it is building and this elite, “Everything we do that’s an elite platform that’s got continuity across the different device subsets,” I think it really shows. It goes to show that Qualcomm has created efficiency and, of course, they’re doing so by continuing to grow, continuing to develop, and being cognizant of how they’re spending on R&D. And that’s another area that the company’s done pretty well.
So this all really became evident here at the event. Pat, I think as we spent some time there and as we know, there are some question marks, great technology, but we are in a particular era of the market where AI PCs, how good are they doing and are they growing? Our intelligence team will actually be launching some forecast data on this over the next couple of weeks. Handsets, we’ve seen the major pullbacks in iPhone 16. And while that doesn’t directly impact Qualcomm that much as their contents fading out of the Apple ecosystem, the question mark about what is driving people to buy handsets and how quickly will this AI cycle really take place is another question mark. But I think when it comes to automotive and you hit this and you really went into detail is, they’ve done just a remarkable job of winning basically all the designs across every OEM and doing so, as you said, in a piece part or full system implementation.
And I love that in our conversation with Nakul Duggal, the group GM of that business and several others within Qualcomm, he really said they’re focused on businesses out there to be had today. So there’s Robotaxi Day, which is great and that’s fun. And then there’s the cars that people are going to buy in the next 12, 24 months. There’s the implementation of ADAS, infotainment, telematics. And Qualcomm seems to be there and the revenue is ramping really quickly. Pat, I will end on saying there is a weird missing element of 5G. I don’t know if this is just because AI is the thing because it’s a distraction because 5G never amounted to what it was expected of it. But it is fascinating that this was the all about 5G thing for about five years and now as we enter this AI era, I don’t think I heard it one time in three days on stage and that was really, really interesting to me.
Patrick Moorhead: Yeah. It certainly remains to be a huge part of their business. And even though Apple’s been going away for seven years ever since Apple bought Intel’s 5G assets. Give a company in the top five valuation enough time, they’re going to figure that out. And by the way, even if Qualcomm over time loses the digital part, good luck getting rid of the RF part. So anyways, let’s move forward. Apple’s 25% of global smartphones and Qualcomm still has access to that other 75%. Hey, let’s dive into ServiceNow earnings. Daniel, was that company able to ride the AI transactional wave?
Daniel Newman: You’ve put out some good tweets this week. I’ve seen a little snark in the Moorhead Twitter wave about agents, about agent advertising, about competition. And there seems to be a battle, a war being fought between CEO Marc Benioff and Salesforce and Microsoft right now, but very quietly or maybe not so quietly anymore. ServiceNow seems to be building a platform and a series of agents and announcing partnership after partnership and just absolutely annihilating growth of enterprise software. And it did it again. So you’re talking about a beating across the board. I had the chance to spend some time with CEO, Bill McDermott, on the day of earnings, the actual number of releases that came out with the earnings was palpable. Partnerships with Rimini, partnerships with numerous big consulting firms, partnerships with Databricks, with Snowflake and a new Nvidia building out generative agents partnership that right now looks very, very promising.
Tons of growth in the company’s large customer base. Over 5 million, over 1 million, 15 deals, over 5 million in ACV, double or up 50% on a year over year. Six deals over 10 million. The company’s deal size is growing palpably and this has a lot to do with the fact that they’re moving from this ITSM company to this, “We do everything that you can do with data and we’re going to do it on the Now platform with agents.” And I think they’re not so quietly a competitor now to everybody and that’s an interesting change. They had a great opportunity to coexist, but isn’t that often how competition rises? They rise right alongside until they get big enough, they get enough cash flow, they get enough growth and then they can start to hammer into areas. So they’ve expanded into HR, digital experiences, CRM, and now they’re really talking about data management.
So Bill talked on the call about this 200 plus billion dollars data management opportunity. And Pat, how much time have we spent talking over the years about the data fabric, the impact of a hybrid data fabric, the opportunity for applications to be able to point at data across different data architectures and types and do so in a way that can enable less abstractions, more answers, more efficiency inside the enterprise. And the opportunity for agents to do what agents can ultimately do for the company, which is navigate and orbit between different enterprise software silos to be able to provide the best generative outcomes and outputs and be able to reason through a multiple layer set of decisions to help companies understand opportunities, understand attrition, understand supply chain complexities, understand service gaps,. And that’s where it’s all heading. Now again, some of this is talk and I wrote that piece this week about apps going away.
I think it’s a little sensational, I even put it in my tweet when I wrote the story. But over time what is going to happen is, Pat, I’ll say it, you can shake your head and disagree with me. No one likes using their app, their enterprise apps. And I’m not saying you disagree, I’m saying you can if you want, but nobody that I know is really like, “I love my experience of trying to get a report out of name tool, Oracle, SAP, Salesforce Dynamics. It’s not like one is great and the rest are bad. It’s like no one really thinks it’s great, but we also know that our entire business sits on top of these data sets and we need to know what’s going on. So the idea that if you like the experiences of generative tools like… Well not Copilot itself, but OpenAI, like Anthropic and their LLMs. And you want to be able to just use multimodal to ask a question, “Hey, show me my next 30 days, my best sales opportunities.” And to be able to navigate across all your platforms, all your data, all your tools to be able to generate in real time a dashboard based on role and permissions and all the sovereignty of data and all that, doing it one time.
That’s where it’s going to go. Whether this is going to get us there, that’s where it has to go. And that’s what I see. Bill is very adamant that this is the company that’s going to get it done. Are they the company that’s going to get done? Look, I let the growth speak itself. It’s hard to argue with this many quarters of growing at two and even three times the enterprise software market, according to our intelligence. So that speaks to a lot of what’s been going on. I think there’s an opportunity and if Microsoft and Salesforce are going to have it out, it could be an interesting opportunity for ServiceNow to quietly just come right through the middle of them as they are having it out. But I also think that these companies all get partnerships, they get ecosystem, Pat. So a lot there, I didn’t hit tons and tons of detail on the earning. The last thing I’ll leave about the earnings is huge RPO, huge backlog, which means the next few quarters look very optimistic for the company because as you know that’s how this works.
Patrick Moorhead: Yeah. Good breakdown here. Sorry about all the coughing here.
Daniel Newman: Yeah.
Patrick Moorhead: I did a 500 at the very end of my workout and I’m pretty freaking gassed. Yeah. It was tough. I did hit above average, but I expected to be superior to my age group. And I did very good, but not as good as I would’ve liked. I also did it at the end of my workout, not at the beginning. Yeah. ServiceNow, I went to their big tent event with you in Las Vegas where we did some selfie analysis with a lot of the senior execs, not only from ServiceNow but also Jensen and Michael Dell, it was pretty awesome. But two big things came out of that. First of all is the declaration that, “We want to be the AI transaction layer on top of these enterprise SaaS apps that have a crappy experience.” And yes they do. They’re absolutely horrible. I’m just going to tell you that right now. And I think when we look at what a good experience is, they are things like on a smartphone, right? These are not like that. And it’s not just putting a better UI layer, but it’s putting the transaction layer so you can actually do something with that.
One very important conversation that began, that again, I had the most questions on, was what’s your data management strategy? I felt a little bit like it caught ServiceNow off guard a bit because of all the questions, but to their credit, within a month of their big tent event they really leaned in and they had, I think, a better articulation of what their data management strategy is and what it needs to be going on forward. Now like you said, we’ve been talking about data management before AI for a long time here. And we also got to give Marc Benioff credit at Salesforce, he hit this super hard and his biggest growth percentage-wise is from his data cloud. Hats off though to ServiceNow on… We’re looking at enterprise SaaS hitting it at single digits right now industry-wide. And now there are pockets like with certain ERP vendors that we might be talking about soon that are having some pretty good growth and we’ve seen some pretty good growth from Oracle ERP, but the rest is not great.
And I think a lot of that is, first of all, some digestion left over from the pandemic, but also budget shifting to AI. And I think that’s hurting, quite frankly, the uplift in the overall market. So let’s move forward here, Daniel, and we’ll go to the next topic because we hate talking about chips. And that is that essentially the day after Qualcomm’s, big announcements Arm declared that it was canceling Qualcomm’s licensing agreement. And let me give a little bit of breakdown here. So Arm is not in the business of making chips, it is in the business of licensing intellectual property and that’s about 50% of their business. So what essentially is an upfront payment for the right to use Arm’s technology in your product development, right? Apple has this, Nvidia has this from the failed takeover bid, Huawei has this, and Qualcomm has this. A company that Qualcomm bought, NUVIA, had a license as well.
And about 50% of Arm’s business are royalties. And these are ongoing payments based on either the sale price of the final products or the chip price. And you typically see these types of structures when you’re buying these macros. Qualcomm is primarily in the business of selling chips, secondarily in the business of selling intellectual property, primarily around 5G. So with the basis of that, Qualcomm bought, again, this company called NUVIA, the ex Apple Bionic architects and designers. And Qualcomm used this capability to create processors for what we saw in the PC market and what we recently saw in the smartphone and the auto market. Arm is arguing the Qualcomm should pay the higher licensing fee. And again, it is funny, I asked so many people, “Hey, which was higher? The NUVIA cost or the Qualcomm cost?” And there’s also been discussions, I know from Arm, that says, “Hey, you might have bought NUVIA, but you do not have the right to distribute their intellectual property.”
So Arm is suing Qualcomm, I believe Qualcomm countersued, and it’s supposed to go to trial on December 16th. So what Arm with the 60 day notice said is essentially, “You have to stop selling any chips with our licensing IP in it within 60 days.” And there’s two ways to look at that and I know that I’ve gotten calls from a lot of people both sides setting the record straight with me. On one angle you have to look at upside. So let’s just say that Arm would win, it’s absolute upside for the company because that incremental licensing revenue is not in their P&L now. Now the opportunity cost if they don’t get that would be higher just based on the size of the revenue and operating income difference between the two companies.
The other way to look at this is that let’s just say that Arm got a win here and they got the ITC to blockade all of the Qualcomm chips. That would mean that Qualcomm couldn’t sell any of its chips with NUVIA-based architecture in it. And this might even extend to all Qualcomm… Sorry. All Arm licensing, which would means it couldn’t ship any of its products that are Arm-based to anybody. Now in my mental model, and I outlined this on Yahoo Finance, is that I find that an ITC blockade is highly unlikely. And the reason is I’ve seen this before. How many times has Apple received an ITC block for stomping on somebody’s intellectual property? I think I counted five. Okay? And how many times did the ITC say, “You have to stop selling,” in there? Once and it was Massimo for Apple Watch sensors. Heck, even the President of the United States intervened when Apple lost the ruling, I think it was Trump, lost the ruling and got an ITC blockade.
So mentally I do not see that as an opportunity because quite frankly it would destroy Qualcomm. Okay? And I don’t think that would be necessarily in the cards. I do believe that these companies will find a happy ending to this. Their customers are not happy, their ecosystem is not happy and we need to get this thing straightened out one way or another. I think, like all negotiations, I think we’ll land somewhere in the middle between what everybody is asking for.
Daniel Newman: Yeah. That was a great breakdown, Pat. I think there’s a lot there and have had a number of conversations. We’ve had chances internally talk to executives across the spectrum. You give the background, what I’ll say is a couple of things that come to mind. One is, you and I have both, and I’ve certainly followed seven or eight different Qualcomm cases, substantial litigation with Qualcomm, FTC, CFTC, JFTC, KFTC, Apple, Samsung, Huawei, the Broadcom attempted takeover, which is a little different than traditional. But Pat, this company is incredibly proficient at dealing, prolific almost, at dealing with these kinds of situations. And so it’s a very interesting moment for Arm and Qualcomm, two companies that have extensive legal foundations, right? Both have big licensing businesses, both probably have almost as many lawyers as engineers, in jest. But in serious, the joke about Qualcomm has been the world’s largest law firm.
Patrick Moorhead: A lot of lawyers.
Daniel Newman: A lot of lawyers. And the companies have done a great job in very tough situations and this seems to be just another hop, skip, and a jump in the life of being Qualcomm. Having said that, on the surface Arm seems to have a very real case to be made here. I think in history Arm has almost always granted permission to utilize license and take over, this one never got approved. It was somewhat cut and dry in the architectural license that NUVIA had that it wasn’t transferable without Arm’s permission. So on the surface it seems like it’s a pretty straightforward thing, but that’s the surface. Problem is, in law it’s never as obvious as what’s on the surface.
And so now you’ve got a battle. And by the way, to your point about ITC blocks and everything else that goes on, you got a battle that impacts a lot more people than Arm and Qualcomm because Arm and Qualcomm feed IP to OEMs and device makers that make products that end up in our hands. So you mentioned 75% of smartphones, Pat, if all of a sudden 75% of the smartphone market and they don’t have all of that 75%, but 75% that they address loses one of the biggest and actually the biggest and is suddenly only, what, MediaTek left to provide? You would’ve a major bottleneck in the production-
Patrick Moorhead: Exactly.
Daniel Newman: … of next generation smartphones. So who loses in the end? The OEMs lose, the resellers lose, the customers lose. And so that’s what, to some extent, makes this almost like a popular opinion battle more than a technology because a judge is going to have to not only rule on the law and the legal basis, but also on the impact that making any sort of what I would call really substantial decision could have on the industries that these companies feed. You’re talking about cars, you’re talking about laptops, you’re talking about phones. And of course you’re talking about a company that depends on royalties, another company that basically creates multiple very large, hundred plus billion dollar valuated companies, most of the equipment or pieces of what the intellectual property they need to build their products.
This becomes fascinating. In the end, I think you hit it on the head, there has to be common ground. I just do not see any situation in which these companies are not brought together to have to fight. Are they negotiating? I couldn’t tell you for sure. Should they be? I think it benefits everyone in the ecosystem that they find that common ground. I’ll end where I started. Qualcomm is very good. They deal with this in stride. You saw very little emotion in this as we were around a lot of the executives this week, it seems like just another day of business. So Pat-
Patrick Moorhead: Yeah.
Daniel Newman: We’ll have watch to closely.
Patrick Moorhead: No, I’m glad you pulled that up, right? It wasn’t a bunch of Qualcomm people running around. It was just literally another day out there.
Daniel Newman: Yeah.
Patrick Moorhead: And I know there’s some discussion about parties not negotiating, which I would just find weird. Maybe this is a lawyer-to-lawyer thing. But yeah. This is a good one. Yeah. I love chip wars. The only thing I like that’s second after chip wars is enterprise SaaS versus agent wars.
Daniel Newman: It only started last week though. I’m joking. What did you tweet about that? Five months ago, six months ago you put out a tweet reviewing that this was about to happen on the agent stuff. That was pretty good.
Patrick Moorhead: Yeah. I was sitting in Google Cloud Next and I’m just like, “Oh, what Google’s not saying is that agents can potentially replace enterprise SaaS,” which got no traction on that. But yeah. Benioff really, really likes Copilot here. Anyways, let’s move to the next topic here. We’re going to do a two-figure here, folks. We’ve got IBM Q3 earnings and IBM introduced some very provocative enterprise LLMs, which I’ll let Daniel talk about first.
Daniel Newman: Yeah. So I’ll start off, we’ll talk a little earnings and then maybe we do the second part.
Patrick Moorhead: Yeah.
Daniel Newman: Let’s talk about the Granite 3 launch because it is a bit of a two-for and there is a bit of a thread to tie it all together, Pat. But you and I had a great opportunity to sit down with CFO, Jim Kavanaugh, on earnings day, get the rundown from the company. It’s sold off pretty hard in the wake of its results. Revenue growth was seen as a little bit slow. Now this is a company that’s historically been valued more on free cashflow than on growth. We know that under the Rometty era it didn’t grow at all most of that time. We know in the Arvind Krishna era it’s been what you would call low and mid single digit growth. This era has been the hybrid cloud and AI. I think we both agree and we’ve said many times that the company’s focus has been better in this era. I think the spin-off of Kyndryl, the removal of certain parts of the business that just weren’t priority, weren’t focus areas for the company has been good.
It’s a strategy of playing hybrid cloud and playing partner to the hyperscalers as opposed to really meaningfully trying to compete has served the company well. But I think there’s been a lot of what I call puts and takes, push and pull with the company where once some part of it starts to grow the way it should, some other parts starts to pull back. It’s got the cyclicality angle to it. So it was a very tidy quarter from an operations and finance standpoint with a continued ability to create cashflow to deliver op inc and percentage op inc has enabled to company to at least show growth and continually hit the EPS results that it’s been promising to the street. The challenge is it’s seen this great growth. You’re talking about a $3 billion order book on AI, about what I think we heard a number of something like 4/5 of that is consulting, the rest of it is software and the Watson family. And this is really encouraging. Having said that, somehow consulting isn’t growing, which is a weird thing. You got all this demand, all this interest and all this need for consultants to be able to deliver and implement generative AI. They’ve got a great platform. I think the first real enterprise end-to-end AI data governance platform.
And yet, like I said, with all this order book, all this demand for consulting the growth of the company is low single digits, the growth of consulting is actually negative. And so this has been the interesting area to sort of navigate is how does the company deal with it having this powerful tailwind with AI, but core technology consulting is down, core hybrid infrastructure is down, Z is late in cycle. And anytime Z gets late in cycle that really hurts the infrastructure number for the company because the good news is there’s a new wave coming. The bad news is at that tail end, the revenue falls off hard and therefore it becomes more and more pressure on the company to make up for that. The only other thing I’ll say is that as we move to AI, it’s very clear that the reason IP budgets are flat is that all the spend has moved to AI and all the other spend is basically just stopped right in its tracks. That’s why you see on that chip side, CPU growth is on data centers down to nothing. And so if you don’t have the ability to augment a lot of revenue over to AI and you’re not getting enough there, you’ll actually see a lot of growth get offset. And that’s really what’s happened here with IBM. There’s no way to split hairs about that.
The good news, Pat, and I’m going to let you comment on earnings and maybe we go to Granite after that, is I do think the technology for the first time really is showing some strong market leadership. We’re seeing that with customer wins, they’re seeing it with use cases, they’re seeing it with outcomes that consulting is driving. It’s one of those situations where all the good things haven’t happened at one time for IBM in a long time. So we’ve seen some good Red Hat results or we see some good AI number results or we see a good consulting result or we see a good Z, almost never all these things in one quarter. And so that’s why it’s all been at that low single digit. But to the credit of the company, it’s dealing well with the fall off of the more legacy businesses. It’s done a good job offloading them. It’s done a good job of offsetting them. And that’s why at least the company continues to grow in the stock while it’s falling hard on this earning. Actually over the last couple of years has been a really good performer because the company’s been so consistent.
Patrick Moorhead: Yeah. Dan, that was a great breakdown. And yeah. I was surprised, you would expect consulting to be off the charts. And then what you do is you look at the industry where Accenture just got back to growth. So I am expecting that IBM will get back to growth with consulting. And I think I asked Jim twice about consulting to break it down. And at the end, after a little bit of research, it made sense for me that those dollars are moving from consulting to doing POCs. You might think that, “Well, doesn’t it take consulting to do a POC?” Not always. And again, at the end I look at Accenture, them just getting back to growth. And I’ll do a side eye if we don’t see some growth next quarter because IBM would not be moving with the market. But let’s look at some of the bright side here, right? AI backlog is up 50%.
Daniel Newman: Oh.
Patrick Moorhead: Red Hat bookings year to day subscriptions up 20% and with that carry stuff like Ansible and everything through it. ARR book on the software up double digits. So while that single digit number is single digits, there were some areas of growth where you would expect that growth to be driven from. The other thing I want to talk about is infrastructure, right? IBM Z is in the longest, longest span of riding the current product line in the history of Z. And we saw it this quarter, right? Infrastructure was down big time. And typically it has been one of the, I don’t want to call it a savior because that’s probably doing it too big, but they were just absolutely crushing it. They’ve been riding current infrastructure for about three years. There is a new platform that’s coming out in the first half `25, which should boost that.
And again, there’s two feeders to downstream revenue for IBM, it’s infrastructure and it’s consulting. And if both of those aren’t hitting in the same quarter, you’re going to get what you are going to get. But hey, Daniel, let’s pivot to the future here and talk about these Granite 3.0 models. So the company came out, made what I think is its biggest announcements in AI, aside from when it first came out with watsonx. So this is the third edition of these Granite 3.0 models. And to be very specific, these are for business, these are not for consumers. First is one called Guardian, and that’s for guardrail capabilities. The other one is an MOE, a mixture of experts, super low latency for CPU deployments and edge computing. You’ve got time series for zero few-shot forecasting, very, very high performance. You’ve got Code Assistant that IBM has had a really good track record.
And the overall comment I want to make here is that obviously by 3.0, these aren’t the first models, but for the first time the company is hitting not only on performance, and that’s accuracy, but it’s also hitting on efficiency. And the reason to believe here that these are better for business. And again, we have not done the testing at Signal65 yet, but I do like the figures coming out from IBM that they look really good is that they don’t test on world data. They don’t test on golf scores, they don’t test on the history of sports. They’re not testing on recipes, vacations, and therefore your data set is smaller and that means it’s going to be higher performance. So for something that might take a hundred B models, they can do with five. And I’m just throwing out numbers that aren’t the numbers because they vary.
And what does a smaller model mean? Is you don’t need a Blackwell to run these models on, you can run them on the CPU, you can run them on NVIDIA L40. You can run them on a lower end AMD, which could make them 10 to 20 times less costly to run your applications. This is IBM’s big move. They need to pour on the gas, on the promotion side, on the selling side, on the collaboration side with their partners like Adobe and with SAP. To make a resounding exclamation point on this. And from a positioning standpoint, the company needs to be very clear. There are models that are optimized for consumer and there are models and services that are optimized for the enterprise. And clearly this is for the enterprise. Good job, IBM.
Daniel Newman: Yeah. Pat, you hit it. And so I’ll just double click on a couple of things. The future for most enterprises will be smaller, dedicated, focused models. Okay? IBM, interestingly enough, has a unique opportunity to be, and I say this very cautiously but openly, a real alternative to the sort of NIMS, NVIDIA, and ecosystem. Okay? The idea of being able to build industry models through the combination of smaller models, through the combination, of having the data AI governance model, of having infrastructure access, and of course being able to take great deep expertise into specific use cases is really the idea. Now, how that gets done can vary from company to company, from technology stack to stack. But where it all starts is having high fidelity, very accurate models that deal with language or deal with a specific use case. And that is something that IBM is doing really well.
And Pat, this stuff changes. It’s incredibly fluid. And so this is a moment in time. The moment in time could change. And as we look at the possibilities and what could change, you need to understand that this is something that’s fluid for the enterprises adopting it. Now, what I will say, and I end here because you covered this so well, is that enterprises move at a different speed than the cloud. Okay? And so this is why I really like IBM and this idea of generative AI for the enterprise, Pat, is that enterprises are going to be looking to solve problems. They’re going to be looking very cautiously at data, governance and compliance are big concerns. Their ability to move at the speed of innovation in the silicon space is going to be slowed.
And so they have a couple of ways to address this, but there’s a ton of need. Our data has shown this endlessly for capability and capacity for specialty expertise. And so what I really liked, and I sat with Rob Thomas last week, I was talking to him, is I like the idea that they can bring together technology and consulting to implementation and do so in a very uniform way. And, Pat, the other thing I like is that IBM is open in the sense of working with other vendors. So I said competitive to NVIDIA, but they could also work alongside NVIDIA. They could work alongside different public clouds and different offerings. They could work alongside different data lakes, if they need to. Because they can make the money in consulting, but they can also make the money…
Now the last thing here, Pat, is just to say IBM’s gap is the market identifying and seeing that its models are in fact really competitive, really valuable, and belong in the conversation. So that’s what I’m saying here is after reviewing this, after looking at the data, after looking at the competitive posture and looking at the power and efficiency and accuracy of these smaller models, enterprises really do have permission to think very seriously about IBM as an alternative to do cost per token reduction with high fidelity to get enterprise outcomes. And so that’s what IBM should be pushing. That’s the opportunity and the challenge. And that’s where I think the company could go.
Patrick Moorhead: Good analysis there, Dan. Looking forward to potentially testing our own.
Daniel Newman: Yeah. We got to get that in the lab, man.
Patrick Moorhead: I would love to get these inside of Signal65. Let’s move to the next topic, which we hate semiconductors, but a big story came out here that the Nvidia Blackwell design flaw was fixed. CEO Jensen Wong came out and said this directly. And he also said that the issue was 100% an Nvidia design issue and it was not a TSMC issue. There was a ton of news cycles on this, Dan, which was big rumors came out that there was an issue. There wasn’t a lot discussed, actually nothing discussed from the company, from Nvidia. It was during their quiet period and need to watch what you do say, but was addressed on the earnings call by Jensen himself and also Collette in the financial note that she put out. And I believe it appeared to be one of the higher-end variants of the platform. And there’s two different variants of this. And these would go to the mega hyperscalers, the highest tier folks.
So that swirled for a little bit. But on their earnings call NVIDIA stuck to their guidance, right? And then over the last couple weeks, I believe it was Jensen himself, saying that the company was sold out for Blackwell for the next 12 months. Again, we follow Neistat, maybe this was a Bloomberg article that discussed that, “Hey, there was this finger pointing that went on between TSMC and NVIDIA.” And by the way, I worked at a chip company, specced the chips I wanted from chip companies, have been doing this for nearly 35 years. And there’s always finger pointing between a design group and the group that actually has to build something. My gosh, we had that at AMD all of the time when we had our own fab and it was like the classic Spider-Man meme, right? They’re all pointing fingers at each other.
But I Jensen popped that balloon saying this was 100% NVIDIA and did not have to do with TSMC. It is wild though. Here’s the deal, there are multiple spins on ships all of the time, right? And typically if you can’t fix it in software, you fix it in firmware. If you can’t fix it in software or firmware, you have to do typically a spin. If it’s a yield issue, you don’t automatically need to do a metal spin, but if it’s a design issue, you typically have to do a metal spin and that’ll set you back 30, 32, 60, sometimes upwards of 90 days. And hopefully you’re not doing the pedal to the metal manufacturing because you’re going to have to grind up all of those chips that you did and start the clock ticking again. But crisis averted, design flaw is fixed. Blackwell is sold out for 12 months.
And by the way, the final thing, it was always our contention that anything over a 60 or a 90-day delay would have no material impact of the company because they would just buy the lower performance chips. There was some question on the lower performing chips prior generation whether you could get your supply chain in gear. There was also discussion about share shift to NVIDIA. But again, if you’ve made the decision on NVIDIA, likely you’re going to go with NVIDIA, the only question is which variant.
Daniel Newman: Yeah. That’s absolutely the case. Pat, I think you really said it well. This is a process that historically just doesn’t get this much attention. It just doesn’t. This is part of the ongoing optimization that goes during each launch and each ramp phase. Constant work to improve yield, constant work to improve performance. And Pat, I think you were the original call out on the mass change that was going on.
Patrick Moorhead: Oh my gosh. In fact, it was live-
Daniel Newman: Yeah.
Patrick Moorhead: … on CNBC with John Fortt and literally I’m reading Colette’s piece on air. Thank you for calling that out.
Daniel Newman: Yeah. Well, look, we may not get Rogan numbers here yet, but there’s some really good stuff in this pod. So hopefully all you that are watching stick with us, share with your friends, especially if you’re into chips and tech. But Pat, this is a fluid process too. And I don’t think a lot of people appreciate the pace of innovation that NVIDIA and now AMD and others are pushing. This work was hard to do on three year cycles.
Patrick Moorhead: Yeah.
Daniel Newman: Doing this on annual cycles is incredible. And by the way, the fix without really having any meaningful hurt to demand was great. I’m reading a TrendForce data report about that says Amazon, Microsoft, Google, Meta, Oracle and CoreWeave have bought all of the GPUs, all the Blackwell that NVIDIA can make for the next several quarters. I don’t know if that’s two, three, or four, but they’re basically saying, “These handful of companies own it, got it.” And NVIDIA is looking like they’re going to be in really great shape. Sorry, bubble bears, your NVIDIA shorts probably are not going to print anytime soon, but enjoy that. Enjoy the struts, maybe you’ll get it right in one of these quarters. By the way, the bears around AI right now, it’s like being the guy playing the don’t pass line. You’re just the asshole at the table right now. Having said that, the big short, the world’s premise has been made on, there are a few people that sometimes get it right when they’re betting against. I think Bill McDermott actually said the other day on CNBC, he basically said, “Anybody that’s betting against AI right now is going to be regret at big time.” I don’t know.
Patrick Moorhead: I’ve been on air like 20 times. How long is this going to last? Infrastructure? 12 to 18 months. Well, why? It’s the same thing. It’s FOMO.
Daniel Newman: Yeah.
Patrick Moorhead: With the hyperscalers who view this as a once in a multi-decade opportunity to go in and get share shift.
Daniel Newman: Yeah.
Patrick Moorhead: Share shift, but also platform shift, right? Moving people around. Everybody views this as a way to get up on the competition, expand the TAM, get into businesses that they’re weak in. And yeah. It’s freaking, it’s FOMO with a caveat, which I said is if you have investors of the hyperscalers calling for action on CapEx, okay? That would be an issue and the only way that would happen is if we got into a recession or back into a recession. So anyways, I’m sticking to my guns. I’m not always right, but I sure am consistent.
Daniel Newman: Yeah.
Patrick Moorhead: But I feel like for the most part I’m right on the stuff that matters.
Daniel Newman: Yeah. Well you get it every once in a while. So the TLDR is crisis averted. Demand is incredible. AI is in a good position. Consumption is still the question mark. Are there other companies besides NVIDIA that are really going to make money on AI? Yes. Is that obvious just yet? No, it’s still a work in progress, but I think we know where that is and where it’s going, Pat. And of course, if NVIDIA can keep getting these hyperscalers to buy 100 plus billion dollars a year every year for a new node, now they’ve got to figure out how to charge enough per hour per GPU to make money. But of course they’re building products on it, they’re training data sets on it. It’s not just rent on these hyperscalers, there’s lots of different ways to make money. Pat, we got one more topic, don’t we?
Patrick Moorhead: Dan, I’m glad you did that. And by the way, catch my breakdown of semiconductors, I’ll be on Power Lunch on CNBC.
Daniel Newman: What day?
Patrick Moorhead: Today.
Daniel Newman: Oh no (beep), I’m on Monday.
Patrick Moorhead: Excellent. How about that?
Daniel Newman: Yeah.
Patrick Moorhead: Well, they couldn’t have you on three times a week, Dan.
Daniel Newman: That’s wonderful. That’s wonderful. I’m glad.
Patrick Moorhead: Yeah. So catch that out. So let’s go into our last topic. SAP cranked out some earnings. Dan, did we see some AI magic? Did we see the cloud backlog grow once again, what’s going on here?
Daniel Newman: Yeah. So that’s been the strength, Pat, of the company. It’s had strong cloud results, it’s had strong cloud growth, cloud backlog. It’s starting to be able to tell a bit more of a story on how it’s going to have AI. It’s restructuring, is starting to move behind it. You’re starting to see how it’s not just pure ERP, but it’s HR, it’s customer experience, it’s got concurrent expense management and the ability to take what it’s doing with AI to take its cloud transformation seemed to be really well received in this particular quarter, Pat. So was it a surprise? Look, they’re on an all time high right now. So is this another AI bubble? Look, I think this has been a company that’s been pretty conservative. I think they’ve partnered a lot on AI, but certainly haven’t overstated where they see AI being part of its business.
They’ve executed well on cost-cutting. They’ve seen, what, 30% growth in their cloud contracts and they’re starting to see AI more as a meaningful. And of course they’re seeing their cloud ERP suite, so 27% growth there. So Pat, across the board, the company’s doing really well. They’re growing operating profit. In fact, just to be clear, 30% of their cloud contracts now are inclusive of AI use cases. So is this the magic AI revenue number in the number? No, it’s not. But what they’re starting to say is that people that are moving to their cloud or expanding their cloud contracts are doing so because of the implementation of AI, which is what people want to hear about. As we’ve talked to the company, and you and I have talked to Christian many times, we’ve talked to others, Pat, is we do believe that they’re doing things very, very strategically. They’re moving methodically. They’re not overstating their capabilities. They’re winning and working hard to keep customers. It’s going to be very competitive with them and Oracle over the next few years.
And of course they have to deal with the same trend we talked about with agents and assistants and how do we deprecate some of the complexities on the front end of these systems and make it more powerful. But where this company’s gold lies is in its data, among the richest data sets on the planet. They’ve got more access to more data for more enterprises in more geographies, and they do so and they’ve always done so in ways that are very responsible, from a data sovereignty standpoint and residency and lineage. So I like the company’s prospects. Pat, this was a pretty straightforward, very successful quarter and congratulations to SAP. It may not always be the most exciting company, but you know what? Right now data is gold and it is the absolute enabler of a future of AI agents and assistants.
Patrick Moorhead: Yeah. Good breakdown there. And I just love this story because there were a lot of questions about the viability of SAP based on how quickly it got in the cloud. And while I do think that SAP was late to the cloud, I also need to recognize that moving anything that if it breaks, can take down your entire business, it takes down your manufacturing, it takes down your billing, it takes down your inventory, it takes down your financial system, you just don’t want to move that. I also want to recognize that there are customers who don’t want to change anything. Okay? And I think based on three different variations, the pathways for customers to get to the cloud, finally, SAP is basically sending its message to it customers, “We’re going to move. We want to move with you, but we also recognize that there are some of you who just won’t go.”
And there’s two ways to deal with this, right? Oracle just cut customers off on a certain date that the time that it took for them to get over was lower, but they cut them off. And actually good to see this transition for the company. Operating profit was up 28%. Cloud backlog up 29%, just absolutely crushing. They raised their outlook for 2024, higher operating profit and free cashflow. The AI stuff is significant, right? In that even the size of these cloud deals, 65% of these cloud deals were over 5 million euro.
And the AI attach rate is actually quite shocking to me, right? Because you have a customer set that seems to be hyper conservative on this, but the ability to put that in there, I think, is super highly impressive. Yeah. Dan, the data is gold, right? Imagine every widget you ever made, right, on the factory floor, every element of distribution you ever had, every single supplier you had, every single customer that paid you a cent and the payment terms having 20 years of history that you can put generative AI on top of to get insights. Talk about skipping a beat, right? Skipping a layer on there. I think it’s just super fascinating and I look forward to see what SAP does on their data platform in the future, right? We’ve seen Benioff with Salesforce lean into it hard. Gosh, he devoted 50% of his time on an earnings called just talk about data.
Daniel Newman: Data cloud, right?
Patrick Moorhead: Exactly. ServiceNow.
Daniel Newman: Raptor.
Patrick Moorhead: Yeah. ServiceNow leaning in hardcore. And I’d like to see 30, 40, 50% of the next earnings call going through everything. That’s data because it’s funny, sales and marketing data is great, but 20, 25 years of manufacturing, finance, distribution, supplier, customers, that is literally solid gold.
Daniel Newman: Yeah. Pat, just one last thing on the way out. There is going to be consolidation and that’s the fight and that’s going to be really exciting. Great for competition, great for use. People will not want to have 27,000 platforms.
Patrick Moorhead: Yeah. Do we want to broach this, Daniel?
Daniel Newman: I can’t now.
Patrick Moorhead: Okay.
Daniel Newman: I do want to do this though. I have a meeting that’s been waiting for me. But yes, for everyone out there, what do you guys think about us doing the agent wars? Maybe having a bit of a breakdown and a battle and a little fun conversation on this show about who’s going to win the AI wars and enterprise software. Pat, how about this, I have an idea, we’re going to release our enterprise apps and software, future of intelligence and data report that’s going to have the market size, category decision data in the next few days. How about I come on and you give me a platform to absolutely just blow my own whistle for another about 30 minutes. I’m kidding. But we’ll use that as a background to talk about why it’s only growing 8% and who are going to be the real winners maybe next week or in one of our episodes upcoming.
Patrick Moorhead: Very provocative, Dan. Very provocative. Hey, next week Dan and I will be out at the Cisco Partner Conference and making a quick stop up at Amazon Web Services to do a meet and greet with CEO, Matt Garman. Looking forward to that. Going to meet with a bunch of press people too. And then, gosh, this is five weeks on the road, baby. I am falling apart. Car parts falling off, engines, everything. But got to stay in there. And also if you get a moment, I talk more chips. I was on a podcast with Scott Galloway, the Prof G podcast, with him and Ed Elson. Had a good discussion on NVIDIA, Intel, the entire semiconductor value chain. I put all those links on my LinkedIn and X. Check Dan and I out this week on our broadcast appearances. Dan has like 27, I think I have two coming up.
Daniel Newman: Actually today, bestie. I think I have seen you on this show before.
Patrick Moorhead: Yeah. I’m sure you just weren’t available or you were on CNBC too many times this week.
Daniel Newman: Ha, no, no, no, no, no. When it comes to what you talk about there’s nobody better than you, except me.
Patrick Moorhead: Ah, bestie. Thank you. That’s so good.
Daniel Newman: I have to go.
Patrick Moorhead: Yeah. Thanks everybody for tuning in. Give us feedback. We really appreciate you. Try to relax this weekend. I know I’m going to try. I don’t do a very good job at it. But back on the road, up in the air at seven on Sunday. Take care, y’all.
