Talking DeepSeek, Microsoft, IBM, Intel, and More

Talking DeepSeek, Microsoft, IBM, Intel, and More

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:

  1. DeepSeek Panic or Joy?
  2. Azure Disappoints While AI Soars
  3. IBM Up 12 Percent
  4. Intel Resets Datacenter Roadmap
  5. ServiceNow Agents-Agents-Agents
  6. SAP Rinse & Repeat

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 Weekly Show is back, episode 247. We are back from Switzerland, everybody has forgotten about Stargate, and we are ready to roll. Daniel, it’s great to see you. It’s been at least, I don’t know, 16, 17 hours since I saw you at The Oracle Event.

Daniel Newman: Yeah, it’s been a while, Pat. It was hard. Remember this weekend after Davos, we actually took a day off before hitting the gym together.

Patrick Moorhead: Yeah.

Daniel Newman: That was a lift-a-thon. I think there’s a video out there of that, that just complete testosterone and rage.

Patrick Moorhead: Totally.

Daniel Newman: It was crazy, but it was a lot of fun. But yeah, we’re back. We came back Monday, big news. Maybe it was news, I don’t even frigging know anymore.
Pat, can I say something before?

Patrick Moorhead: Absolutely.

Daniel Newman: Is this maybe the last week of The Six Five like this? Is something happening on The Six Five? I think we’re getting close, aren’t we?

Patrick Moorhead: Yeah, we’re getting close to a new format. I’m sure you’re going to like it. We’re getting a little bit more serious about our weekly show here. We’re going to have different sections of analysis. We may or may not be doing a point-counterpoint. Yeah, looking forward to it. After 247 episodes, we need to change the format up and I’m looking forward to it. But hey, we do have an amazing show for you today. It’s amazing how many big things come out. We’re out at World Economic Forum in Davos, and Stargate hits, and then the whole world is on Stargate. Then DeepSeek hits, and then everybody’s all over that, going crazy. We forgot about DeepSeek. Then all the stuff going on with the new Trump Administration. Hey, we’re going to bring with you what we think were the biggest conversations of the week. We’re going to be talking about DeepSeek. Is it time to panic or is it just joy inhalation for everybody in the markets? We’re going to talk about Microsoft, how it did in earnings. A lot of focus on Azure. AI soars, but the market a little bit concerned with that Azure growth, that puny 3X Azure growth. Patrick laughs.

Daniel Newman: 3X, that puny 3X Azure AI growth.

Patrick Moorhead: 3X percent. I forget if it was 31, 32. But anyways, it’s big. IBM up 12%! The absolute crusher of the week so far in tech. What the heck happened to IBM? Intel had their earnings yesterday. We’re going to talk a little bit of how they did, but they also reset their data center roadmap on their data center AI GPU in a little push, push on their mini E-core product out there. ServiceNow, one of those companies that can benefit the most of the downstream impacts, actually will make the downstream impacts happen. How’d they do? They were just on a beat, beat, beat, beat, raise, beat, beat, raise, beat, beat, raise. How’d they do? They also had some major announcements as well.

Daniel Newman: Did they beat, beat, and raise?

Patrick Moorhead: Probably not, but we’ll see.

Daniel Newman: All right, all right.

Patrick Moorhead: SAP, what came out of their earnings result? Is it more cloud and more AI? Was it rinse-and-repeat from the prior quarters? We’re going to get into that.
Dan, let’s go. I am calling your number, bestie. DeepSeek, should we all be panicking and keep selling NVIDIA and everybody related to NVIDIA? Or is this an amazing time and joyous time, we’re going to get cheap reasoning and therefore, the downstream benefits are getting closer?

Daniel Newman: Oh my gosh, Pat. Good God, was this week a disaster for the media plaudits, pundits lauding this company. When did we begin to just take China at its word for everything that it does? But like I said, let me try to be well-rounded here. Before I sound like a crazy conspiracy theorist, let me just come out and say that there was some value in what happened this week. The number one thing that I saw was the idea of finding ways to accelerate scaling laws, especially as it relates to inference, making inference less expensive for LLMs to be able to deliver tokens is something everybody wants and needs. It’s too much cost of training. This capex expense is not making sense for investors because you’re looking at the inputs and the outputs, and to spend 80 billion to generation 13 billion is going to cause a sell-off.

Having said that, we are so early in adoption of AI, we are so early. People are like, “We’re far along.” We’re not far along. The game hasn’t started yet. Companies are still mostly in POCs. They’re using this in small vacuums. They’re using small numbers of tokens. Inference is the game. 20-time bigger opportunity, maybe even larger than 20-times than training. By the way, this infrastructure that everyone suddenly thought we wouldn’t need anymore, you still need it to do all this inference. Here’s the difference, is when you’re running agentic AI for an enterprise with 30,000 people that’s going to have three million agents working concurrently across 20 different systems, you’re going to have enterprises creating trillions of tokens per day. Even if those tokens are fractions of pennies or pennies, we’re going to need to go faster with more powerful infrastructure that scales better, that is able to handle the concurrent requests. Therefore, the idea that we need to slow down is wrong. We saw Satya, people like you, me, all talking about Jevons paradox. The thing is is if we can get more efficient in our training and more efficient in our model development, and more efficient in inferencing, we will actually see the adoption of AI accelerate. The way I look at it, Pat, is really simple. Very, very simple. More efficient models means better quality models. It means more innovation, better quality models, and better quality models means faster adoption. Faster adoption means massive scale of AI, which means revenue.

The big thing with AI is it’s been too concentrated so far. Only a few companies are really able to meaningfully show growth, revenue, and profits because of AI. As we scale inference, bring it out to the software companies, service providers, and ultimately to the industries, Pat, we get actual economic efficiencies, growth, and profits that go into retail, it goes into pharmaceuticals. It goes into transportation industries. It goes into manufacturing. That’s where we want to get to. That was the good part. Everybody agrees we want that. The bad part is they had a billion dollars of chips. They told us it was $6 million. This was literally, we took a trillion out of the economy, a trillion-plus dollars out of economy, it was the best short in history. Better than housing shorts, the big shorts. It was better than anything Hindenburg did in its entire existence in one trade. The only question that I’m going to leave you with here, Pat, because we could talk about this one all day, is that I am really hoping that somebody actually follows the money here. What happened here? Because remember, DeepSeek was a hedge fund that actually put this thing together. Whether this is true or not, if there’s a conspiracy or not, imagine actually building this product, writing that paper, and then putting a short right before you actually put this out, and then creating this massive economic boon. Not saying it’s true, I’m just saying that a lot of things that people thought were crazy and movie-like over the last few years have turned out to be kind of true, so call me crazy, I’ll take the roll. But Pat, this was mostly BS with a little sprinkling of enthusiasm about the future of accelerating AI and scaling laws.

Patrick Moorhead: Dan, that was a great breakdown. You hit pretty much everything. Let me try to just fill in the gaps maybe. There’s three things you have to do when it comes to AI. There’s pre-training, there’s training, and then there’s inference. What you have is innovations at every step of the game. We’ve seen this with Llama, we’ve seen this with OpenAI. We have seen this with Gemini. We have seen the API costs and access plummet over time. I think the thing that made this special and pronounced was two things. First of all, it was right on the heels of a $500 billion Stargate announcement for AI infrastructure. Secondly, it came out of China. I would say there’s probably a third thing, is that typically, these cost reductions and these innovations that were coming out of China might be multiple months. But reasoning is relatively new. It wasn’t right on top of the reasoning models from the other companies, it was behind. But it just wasn’t as behind. That got everybody excited about this.

But there were at least 10 unique innovations across that workflow that DeepSeek implemented that others didn’t. They found ways to train and infer on a smaller dataset. They found ways to get closer to the metal when it came to some of these processes. It did make a lot of sense. Now technically, that six-point-whatever million dollars took a GPU cost per hour across tokens across one area of this. I don’t know if these folks wantonly did this or not. I don’t think it matters. But I thought that the impact was good and the conversation was good. I don’t hold any of these stocks, so I’m not crying in my cereal here. But I did think that Net-net, this is good. The only way the train keeps going is if the downstream impacts are realized by consumers, businesses, and governments. It’s up to companies like Microsoft, SAP, ServiceNow, Adobe, the entire app stacks to make that happen. Obviously, Microsoft, Google Cloud, folks like that. Is it panic or joy? Yeah, there was panic. There doesn’t seem to be a whole lot of joy, but I think there should be. I think this is a good thing and it shows innovation.

The one thing that gets lost in this as well, in addition to the downstream impacts, I think is that literally we have so far to go. You said we’re not even on the field yet. Until we get to AGI in all the different types of models, text, reasoning, voice, audio, those put together, and we have the infrastructure on the edge to take full advantage of that, then we can start having those serious conversations. Final note. One of the dark horses here are the end user compute folks. That’s the PCs and it’s the smartphone, and obviously IOT in cars. But if you look at the Intels, the Qualcomms, the AMDs, the Dells, the HPs, the Lenovos, and the Apples, and the Samsungs, at some point, if you can run what used to be a teeny-tiny little model but get the impact from what used to be a gigantic model … There are already people running DeepSeek R1, I’ve seen on AMD, Qualcomm, and Intel silicon. Microsoft made an announcement, I think yesterday or the day before, that they were going to integrate DeepSeek into their developer platform for Windows PCs and CoPilot Plus PCs. I’m excited. It’s a time of joy. But I didn’t lose $10 million.

Daniel Newman: Let me just throw one more quick thing out there, Pat. You and I have long talked about the scaling laws with things like small language models. With the shift from Nvidia chips for everything to accelerators that have actually been, for a long time, creating more efficient for high volume inference for other use cases. We’ve seen training move from NVIDIA to TPUs. We’ve seen specialized training chips coming out of the likes of AWS, and the Marvell’s and Broadcom’s partnering. My point here is that some of this is funny because it wasn’t novel. I guess the idea of bringing the cost of compute down should always be the goal. It’s always been the goal. It’s always been how the evolution, whether it was Moore’s or now what we think has accelerated beyond Moore’s law, Pat, this is all here. One last thing is look, there is a lot of reasons for China to try to slow down the US. Geopolitical and macroeconomic things are real. There is a benefit there. I know, like I said, it can be a little out there, but AI will be the decider of the future of the world’s economy. The fact that we have the newest chips, we have the access to the equipment, and the most innovative companies in the world is an advantage to us.

Patrick Moorhead: Agreed.

Daniel Newman: I’ve said it all.

Patrick Moorhead: All right. All right. This is the topic we would probably spend the most time on, so let’s dive in. Microsoft had their earnings. You know, it was mixed. They banged out a beat and a beat. But people were concerned about that measly 30% growth in Azure, which I just find a bit comical. But listen, if I look down the line of how these folks did, up 14% on productivity and business processes, and I look across the line, office, commercial, consumer, LinkedIn, and Dynamics up across the board. Investors were happy with Intelligent Cloud, which is a conglomeration of everything that they do in the cloud. That includes Azure, GitHub, Nuance, SQL Server, stuff like that. That was up 19%. Think about this. This is nearly a $26 billion business. A couple points that, just to put this into macro. You live by the sword and you die by the sword. Microsoft was getting a tremendous amount of lift through the OpenAI stuff. If you look at right there, their AI life on that entire cloud number, that was a $13 billion run-rate, annual revenue run-rate. Which is just absolutely phenomenal.

But very similar to the conversation we had about the macro view of DeepSeek, I can’t imagine a company that can benefit downstream more than Microsoft on this. This company is a software company. It’s getting credit right now for its infrastructure, which is great. But it’s a software company and it operates in multiple modalities. You want to build an application? Hey, we got you. You want to run an application? Heck, you can still buy Office without having a subscription. Do you want enterprise SaaS through Dynamics 365? Or through the entire Microsoft 365 stack? And let’s not forget GitHub. If there’s a company that’s going to take advantage of the downstream impacts of AI, it’s going to be Microsoft. This intersection upfront that you normally wouldn’t expect from Microsoft related to the OpenAI surge is why people want more.

Daniel Newman: Yeah. Good talking points there, Pat. Look, it was very simple. It’s sold on the Azure growth number, the number wasn’t as high as people wanted. I think we covered a little bit when people started to digest the AI number. The AI, the acceleration of the number. The calculus that’s being done by most people is the 60, 80-plus billion dollars that’s being spent on AI, and knowing that you’ve got a fairly short shelf life on the Hopper, Blackwell. Although, I think there’s a lot of debate on that, by the way, how much shelf life you’re going to have in terms of inference tasks and stuff that can be done longer term. But having said that, to get to 13 billion of revenue. The nice thing for a company like Microsoft, to your point, is really inference can be built into every interaction with every piece of software across every single device. When you start to look at the size of the install base, you can’t really look at Microsoft much differently than you look at Apple. People love to look at Apple for, I know it’s a bit more of a consumer brand, but Microsoft lives inside … Almost every person that works in a company in any sort of role that has access and interaction with a device is using some Microsoft. It’s a massive install base out there.

When you start to think about the ability for them to scale inference and basically deliver it at a margin … Basically every time someone clicks the button, the CoPilot button or any other AI-enhanced feature, it’ll cost Microsoft a penny and they’ll charge three, or whatever it is. Then you start thinking about how this works when you make it trillions of these interactions happening at a time, the math tends to work out. The problem is patience isn’t really there with people. They see the spend, they want a return right away. In the old days, you’d build a factory and five years later, the factory starts creating profit for a company. Nowadays, when you spend that capex, investors want to see returns come really quickly. Microsoft’s growth’s a little bit slower. You’re talking about a fairly muted guidance. They’re 10% growth. Better than Apple. But anyways, the point is, the growth of the whole company and the diversified portfolio, everything from Surface devices, PCs, Windows licensing, to Dynamic software, to of course Azure. I really do think though, Pat, that the companies that have hyperscale cloud, they trade on their cloud number. Everything else is a wash, as long as the cloud is growing fast enough. Just a reiteration about the value of what’s seen as recurring revenue, subscription services, and of course growth of those business areas.

But I’m still very optimistic about Microsoft. I appreciate how transparent they’ve been with their AI business and how it’s generating revenue. I think they’re a company that, by the way, has a great example of a beneficiary of scaling laws. If they can deliver inference cheaper, every one of those interactions with every piece of software and hardware becomes more profitable for the company and/or it drives speed of adoption and enables Microsoft to be more aggressive with how it incorporates inference and how it pushes adoption across all of its softwares that are, like I said, used by so many people. Pretty good numbers, not great. The market’s sold hard on it. It was all about missing that 31%. People want to see the cloud infinitely growing at an incredible pace, and they want to see obviously AI accelerating at a pace that’s probably unrealistic until we see this infrastructure build out and we see some of these scaling laws better implemented into the enterprise workload.

Patrick Moorhead: Yeah, good insights in there, Dan.

Daniel Newman: Thanks.

Patrick Moorhead: Let’s move onto IBM. What the heck? IBM was up 12% the day after earnings. For the week, they’re up 15%. What on Earth is going on here?

Daniel Newman: Was that NVIDIA, or was that Tesla, or was that IBM?

Patrick Moorhead: No, this is fricking IBM. What did they talk about and why does it matter?

Daniel Newman: Well, Pat, first of all, the God candle. Did you see my post? It had the God candle.

Patrick Moorhead: Yeah. How many views did you have on that on LinkedIn?

Daniel Newman: It had the God candle. It was like, “Oh my gosh.” Pat, you and I looked at the numbers. I think it’s always this weird digestion with them. You’re like, “Well, it grew 100% faster than it’s been growing.” It was at 1%, now it’s at two. You’re like, “Oh, that’s great.”

Patrick Moorhead: Yeah.

Daniel Newman: Both you and I had a chance to talk to CFO Jim Kavanaugh. This is a company that does trade on cashflow. It’s a value stock, it’s a cashflow. It’s really, really well managed. The company has been able to genuinely push its software number up from 25 to close to 45% of its revenue. That of course is going to help see it trade at a higher valuation. It’s creating and putting off a lot of cashflow. It’s generated a zero to $5 billion AI business over six or seven quarters. It’s growing that at a fairly high sequential rate. The consulting business has basically turned over, so it’s gone from traditional IT consulting now to almost a pure AI consulting business, which is the pivot that needed to be made. It actually performed pretty well, even given this weight that’s put on it by a late cycle infrastructure business, because the Z business is so cyclical to the company. When it gets late cycle, that business tends to shrink, and shrink, and shrink. But it’s coming into a new cycle, so if you’re looking.

Pat, here’s the money shot. I know, I know. 5% growth estimate next year. Now again, some people would be like, “5%?” That is four-times faster than IBM’s been consistently growing. There’s an optimism that they’re turning a corner, that this AI investment consulting pivot, that this next phase of the Z mainframes, it’s going to all come together in 2025 and it’s going to start to drive growth that’s been much faster. The amount of cashflow the company puts off at that growth is going to be really, really impressive, putting the company in a good position. You also sense that, with deregulation and the Trump Administration, this is going to be a company that’s going to always capitalize in M&A. They’re going to be moving quickly, investing hard and get into growth cap. But look, it was trading like a hyper-growth stock for that day. Congratulations to the longs. A lot of people, probably my mom and dad, and your uncles and aunts that own the stock and they’re taking the small dividend. This week, on Wednesday, Tuesday, whatever day it came out. This week on Wednesday, everybody made money.

Patrick Moorhead: Yeah, it was a good thing to see. I always like that you reset everybody, that this stock trades as a cashflow stock. It is funny, though. I do look at Apple and just wonder is Apple really more like IBM? They’re safe. There’s not going to be some wild gyrations left-and-right. Some of the additive thoughts here is I was really impressed with their generative AI orders. It went up $2 billion sequentially. That’s one of the biggest numbers that we hear. First of all, it’s hard to get that number out and it’s hard to raise that number. Overall, since July of ’23, five billion in aggregates. You can imagine the curve on that is moving there. Jim Kavanaugh, when we were talking to him, I tried to ask him, “Hey, is this the beginning of an absolute rocket ship here?” He didn’t bite on that at all. But I felt like the impression was it is going to be up, up and to the right, that wasn’t some spiky spike. To your point on software as a percentage, 45% of revenue. That’s up from, I don’t know, 22% in 2019. Here, Arvind really did what he said he was going to do. When he took the helm he said, “I’m going to turn my company into a product company.” Arvind got the Red Hat deal over the line, and I do believe that this was his idea, not he inherited something. The company also spun off some non-growth services, got those out there. IBM is a product company, not just a services company, which was a look to that. Back to the growth, 2% doesn’t swing everybody around the room, but it was an overall good showing. I do find it amazing there was any growth without Z. Z is getting super long in tooth, I think it’s in the 11th quarter, and looking forward to any future announcements that they have.

Anyways, let’s move on, Dan. Intel had their earnings, and I think everybody was waiting for bated breath to hear what the new co-CEOs had to say. This is a natural thing. Whenever you put a new leader in place, there are going to be changes or a potential reset, and that includes even when you have the way that the board has positioned these leaderships as temporary. That doesn’t mean they aren’t being considered, but MJ and David are not the official co-CEOs in perpetuity. With that said, just a ton of takeaways from the call. Well, first of all, they had a beat and a beat, so things were elated. I was really surprised after the conference call how positive everybody was still on the stock. I thought it was going to tank after the following. Which was Falcon Shores was essentially canceled, and that’s their low-precision data center AI GPU. They did Telegraph this so maybe that incorporated it, or the street just had zero expectations anyways. But the focus then becomes the next product, which is Jaguar Shores.

The other thing that I thought the street might react to was an on-paper push out of Clearwater Forest, which is the many, many, many E-core that was supposed to, was discussed at getting out in the end of ’25, but was reset to the first half of ’26. Clearwater Forest, to me there were no surprises. Through my channel checks and relationships that I have, they were telling me that it was going to be ’26 not ’25. As MJ reported in, I think it might have been in the Barclays’ or Bank of America conference in December, this is not a great part. I think this is the right thing to do. This is the time where you reset your roadmaps, you reset everybody’s expectation, and you really set the tone for what the future holds. MJ was very insistent on the call about truing up this roadmap and delivery schedules. It’s funny, a lot of people forget and maybe just because it was too long ago, but Intel had a crisis when AMD was pounding it into the ground over a decade ago. When Intel came in with the Core architecture, they reset the roadmaps. I think the wording was, “Hey, we’re moving our roadmap confidence from 80% to 95%,” and they pushed out pretty much everything. None of this shocks me at all.

The other thing on the call that came up on the Foundry side is I really like the pragmatism that came out. 18A might be really good, but that doesn’t mean you’re going to have a ton of customers bet the farm on it. That’s just not going to happen. As we’ve seen, Microsoft and Amazon have signed up, and they’re doing networking chips. They’re not doing core compute yet. What Dave Zinsner talked about on the call was, “Hey, you got to have this trust in.” I think it was one of the first times I had heard that type of pragmatism. I thought it worked. I do wish that Dave would have reported, or MJ, the exact state of 18A with yield numbers or making a comparison to prior nodes, or maybe even in comparison to what TSMC is talking about, too. But other than that, I thought it was a reset call, it was required. I was impressed with the pair.

Daniel Newman: Yeah. I thought it was a really positive call. I commend Michelle for her tact in the whole thing. This is definitely a difficult situation. And David, too, of course. David’s always been … He’s very calm, CFO-like. That’s his demeanor. MJ is a passionate, product and revenue leader of the company. Of course, sometimes patience is hard. You hit a lot of the things that are important. I guess I’d just make two points that really were clear for me. One is all the basically the value of this company has already been soaked out of the market. There’s not a lot of downside left. I think trades are close to book value. People basically here, to see it meaningfully go down from here, you actually have to believe it’s realistic the company’s going to go bankrupt. That’s the bottom line. There isn’t a lot of bad that could be said that would send the stock down, unless the company’s risk of going concern that it could literally go out of business.

The side of it that is good is Intel’s probably fault over the last few years, because I still don’t think Pat got the right treatment on the way out. My personal opinion was he was doing a lot of what needed to be done, he was still stitching a lot of wounds from the past. Could he have put more energy into a low-precision data center GPU and maybe gotten there sooner? Maybe. The point of MJ’s comments was what they had just wasn’t that good. That’s my interpretation of her comments, is the reason they’re going to skip pushing out even another year was because the only thing worse than not having an AI product for the data center is having a censored one. She’s basically saying, “We’re going to get something that’s going to be competitive, it’s going to be right, it’s going to fit the market. We’ve got to get software right.”

We heard from Mark Papermaster at the Oracle event this week, talking about how important getting software right is in this year. You can have a really are GPU, and actually they have a really good accelerator with Gaudi from a performance standpoint, but the software is pretty inflexible. So the customers that were highest probability to consume a lot of Gaudi were the hyperscaled cloud providers because they’re the ones that have enough volume on a workload that they could basically deal with the software. But they all went ahead and built their own. That created a lot of friction in this market. Nobody is going to be satisfied with the DCAI business until they get that done. But now it’s out there. It’s out there. Look, it’s Gaudi from now to ’27 or whoever is going to lead product for that company and CEO of that company long term has to get a good AI product out.

The other thing I’d say, Pat, and you and I have talked a lot about this. In the PC and client-side, the company’s done a good job showing at business. That has not been a bad thing. They’ve done a good job of protecting market share, the Lunar Lake product was good. The roadmap is encouraging. If they can keep the market share they have, and they can protect the business, and they can get to these next generations and build on 18A, and show that that’s an improvement, they’re well-positioned. The last thing I’d say, with all this China noise, there is another loop back of where Intel could really play a part in de-risking the situation of what’s going on with China, with Taiwan, with the lack of US manufacturing of the most advanced chips. I think you said in a segment, even if TFMC brings their leading nodes here, they don’t bring the IP. If the US wants to make sure it’s got that resilience, Intel, all roads still kind of lead to Intel at this point. However they can get that done, that would be a huge benefit.

Patrick Moorhead: One final comment I thought was interesting, not a lot of detail and not even questions from the peanut gallery, was there was a comment. “MJ, she’s got enthusiasm for building both semi-custom and custom products.” That went with a thud. I thought there would be a lot more questions. Two types of ways you can do a custom chip for somebody. You can do custom vids and fids, which is essentially voltage and frequency, to manage the price, performance, and power equation. The other thing you can do is you can actually build a straight-up, it’s funny, it’s called semi-custom, but it’s really a custom chip. AMD does this a lot for Xbox and for Sony. Nobody bit on that. My thoughts go to the following, where Intel has gone to a chiplet architecture, and wondering if they’re going to invoke chiplets and start fricking selling chiplets to, I don’t know, AWS, to Meta, and do a custom processor for them. Then I’m wondering, what does that mean to AMD and what does that potentially mean to ARM? I wish there would have been a question, but there wasn’t. Maybe we’ll follow up with MJ afterwards, I don’t know. Here we go. We’re ahead of schedule, buddy. We’re going to get you to your big company meeting on time.

Daniel Newman: Corporate meetings, my favorite thing to do.

Patrick Moorhead: Corporate meetings, that’s right. You’re a big company now, Dan. ServiceNow. Super interesting, interesting company. Not only did they try to improve the experience of companies like SAP and Salesforce and put a transaction model on top of that, they’re also, with AI, pulling together those applications across, and really becoming the AI agent transaction. How’d they do in earnings? Talk to me about the, I don’t know, three announcements they made on earnings day.

Daniel Newman: Yeah. There was a lot going on, Pat. First of all, just numerically speaking, good quarter. A tidy quarter was how I characterized it. Bill McDermott is an impassioned sales leader. I spoke to him before earnings. He’s incredibly excited, surprisingly, about agentic AI. I don’t know if we’ve heard anything about agentic, but the company’s working on orchestration, and really out-of-the-box agents that will create incredible scale. They’ve got some good early customer wins. They’re really trying to build this to be fast to implement, fast to deploy. The company is moving quickly from really the back of office, the IT and systems, automation and software company to front-of-office with CRM and DX. The Raptor database adoption is growing, and now they’ve got the agent play. That’s all very, very encouraging. And in the numbers it’s encouraging. The large customers, the RPO numbers, the 21% growth number. But they didn’t guide as high as people wanted, and that did lead to a sell off. This is a company that, I think at least twice last year, beat and raised, then beat the raise. But in this case, the guide was more conservative. It had a little bit to do with FX. It had a little bit to do with some of what’s seen as headwinds, some of what’s seen as cost to build out all of this scale and agentic infrastructure. I have a feeling it was a little bit of a, “Hey, we’ve had a great run, the stock is up, customers are happy, they’re growing very quickly. Let’s reset the market a little bit.” I will be very surprised if they don’t beat, just based on the backlog, the RPO and the customers. This is becoming increasingly a very predictable business. I think he could very simply figure out how to get to those numbers.

The company also did a $3 billion share repurchase. Just like Apple, everybody’s got to engineer a little bit, take the float down. But they also announced some great partnerships. With Oracle. They announced a partnership with Google Cloud. Now they’ve scaled out their cloud partnerships with Visa. Like I said, they’ve built this orchestration layer for agents. Pat, I think this is really, really important, is going to be the ability to orchestrate agents across many, many platforms all at the same time. They’re moving very quickly. They seem to be on the right trajectory. This is a company also that builds a lot of its own infrastructure, by the way. There are some definite efficiencies and economy of scale opportunity now to actually build and deliver your own, and not necessarily pay those high costs, especially if we can identify and implement some of these scaling laws. They did say that they had 150%, by the way, growth in their Now Assist deals that are basically gen AI net new ACV. They’re not giving quite the dollar numbers, but they’re showing some very strong growth inside of that area of the business as well.

Patrick Moorhead: Yeah. I’m always impressed that that RPO number, basically this is what’s been sold that hasn’t found its way to revenue yet. It’s $22 billion. That’s a 23% increase. I said a little bit in the run-up, but what I find fascinating about the company is they’ve got their own specific apps doing certain things. Then they can sit on top of other people’s applications, like a Workday, SAP, Salesforce, and to be able to make that more transactional or improve it. My daughters work at Dell, where ServiceNow is sitting right on top of Workday for onboarding and stuff like that. Then you’ve got this third element, which is okay, I’ve got my ERP, my SCM. I’ve got my CRM system, my HRM system. How do I pull the value out of all four or five of those systems? That’s where I find the most fascinating opportunity for the company. Now to do that, data management is an issue. There isn’t a CIO that I talk to that still doesn’t say, “Data management, hey, we got that covered.” In fact, I met with a Fortune 50 CIO last month and they essentially said, “Yeah, we think it’s going to take us five years to get there, before we can have a data plane or data fabric that we can pick up.” I’m like, “Do you have 27 different SAP instances?” They’re like, “No, we have one.” But even SAP, as an example, it’s so hard to get value out of that. You talked a little bit about Raptor, Raptor DB. This is an AI-optimized database, relatively new, brought out last year. They needed this capability two years ago. I know it’s easy to look in the rearview mirror, but I don’t know if somebody was asleep at the switch or they felt comfortable leveraging other companies, let’s say like MongoDB, or being able to tap into Oracle to create that AI-leveragable data play, but I’m glad it’s here now. I’m glad they’re talking about it a lot. I remember at their big tent event we attended in Las Vegas, I know a lot of us had been giving feedback on the need for this capability and I am glad that it is here.

Let’s move on. The finale, the grand finale. SAP had their earnings. On the week, SAP is up 1%. Didn’t get a whole lot of action based on their earnings. But I like to view this as rinse-and-repeat. If I look at the past two or three quarters, it was, “We crushed it on cloud. We’ve got all this AI goodness coming out.” I’m not saying there wasn’t something new, but it was pretty darn consistent. Cloud was up 25% year-over-year. While revenue growth was up still 10% year-on-year is pretty darn good, Dan. Particularly when it comes to these types of software only companies. There were some people that dinged it on profitability, but there’s a lot of restructuring charges going on. Unlike the United States, SAP has what’s called IFRS, which is harder to get credit for your write-offs. In the US stock markets, you can basically bury it. It goes below the line as an extraordinary expense. But in IFRS land, it is not. One thing I’ve been impressed with, shockingly … Not shockingly, that’s a bad thing to say. Was just how impressive their AI examples and use cases and customers. There are real customers using AI across the portfolio, and there are a lot of them. The add-on for AI is in the deep double-digits when it comes to all of the major deals that go in there. Net-net, SAP was late to cloud and it’s making up for it in spades, given the gigantic growth numbers. 25% year-on-year growth in cloud, that’s been very consistent. ERP Suite up 33%. I do not think that they showed up late to the party with AI, but they need to get everybody on the new base of code in order to take advantage of that, and that is their big challenge. Getting them to the clean core, and then initiating AI. It’s funny. I’ve heard so much feedback on how hard it is to access the data inside of SAP to activate. That’s when you’re doing it yourself. But it seems like they’ve got more and more customers every day leveraging SAP to do AI with Jewel.

Daniel Newman: Yeah. You covered a lot of ground, Pat. I think they’re hitting the right growth in the right areas. Where do you want to see growth? The core business is fairly well established. The risk factor of the company is the porting over from core prem-based to cloud. It’s because the situation with the way SAP works is that porting does open up the door for people to migrate off of SAP. They’ve done a lot to invest in systems and services, and things like RISE, where they’re trying to enable companies to move from older to newer. They’re going to ask people to simplify instantiations because there’s so much customization. The customization was always a bit of a lock-down for how quickly someone could move and port. Having to go to the cloud was a bit of a reset, but at the same time, a reset that a company like SAP needed. At 27% cloud revenue growth in the quarter, that’s good. That’s almost as much as Azure growth of cloud. That’s not a bad number. Backlog of cloud, 40% growth over the year. Great. We’re talking about SAP here. We’re not talking about .

Patrick Moorhead: Exactly! Exactly.

Daniel Newman: You’ve got the current, in the quarter, you saw 29%, Pat. You saw cloud review for the full year go up by 25%. These are all really encouraging numbers. This is not something to look down about. This is saying that the work Christian and his team are doing is going in the right direction. They’re locking in customers, they’re moving them to recurring, they’re putting theme into the cloud. If they can do those things effectively, they can get to AI and get to start monetizing the inference, stuff I talked about throughout this entire conversation, Pat. Look, it was a good quarter. SAP is a player. They’re in the space. This migration from the first network effects of infrastructure to the platform and software and services layer is a big opportunity for the company. Like I said, a company of its maturity, to look at it against the Oracle’s, look at it against Salesforce’s, look at it against IBM’s, look at it against ServiceNow’s, it’s in double-digit growth. It’s right in the range. Meaning it’s not going to grow as fast as a high, fast rule of 54 company like ServiceNow, but it’s growing just about as fast as Salesforce. Just remember that. It’s in the game, it seems to be on the right trajectory.

Patrick Moorhead: Great stuff, Daniel. Hey, great show, folks. Appreciate you tuning in. Dan, when’s the next trip we’re going to be in the same place?

Daniel Newman: Together? I don’t know.

Patrick Moorhead: Yeah.

Daniel Newman: Sadly, I don’t know that we have … Is it weird? Is it going to be all the way out to Mobile World before we actually go to the same place at the same time?

Patrick Moorhead: Oh, we need to do a –

Daniel Newman: We have the whole February month apart? Is it the saddest month ever for the Pat and Dan Variety Show?

Patrick Moorhead: I think I just recovered today from that Euro trip.

Daniel Newman: Yeah, I was going to bed at 8:30 every night, and I was waking up at 4:00 AM for a week.

Patrick Moorhead: Yeah.

Daniel Newman: It’s definitely … Man, you’ve got me eating better, you’ve got me lifting harder. I’m in a cutting phase, I’m leaning down now. Those videos are going to be out there. Pat, you’re going to do the shirtless pic next week for everybody out there, Pat’s going to show his six-pack.

Patrick Moorhead: It’s coming. It’s coming, folks. I’m getting the courage. My family-

Daniel Newman: I’m going to wear a T-shirt, by the way, next to you. You know the T-shirt that has all the muscles drawn on it? I’m going to wear the T-shirt so that I can do a shirtless pic, too.

Patrick Moorhead: I don’t know, man. You’re getting a little lean. What’s going on, Dan?

Daniel Newman: I’m going to surprise some people.

Patrick Moorhead: Are you trying to get skinny or something?

Daniel Newman: No, man. I’m taking so much creatine. That what I do is I usually make a drink in the morning and then I mix in some peanut butter, and then I dip it in creatine. Oh, God. That video was so good.

Patrick Moorhead: Yeah. All right, man.

Daniel Newman: I’ll let you send us home.

Patrick Moorhead: Have a good corporate meeting. Hey, thanks for tuning in everybody, episode 247. At some point, we are going to flip this to a new format that I think you guys are going to love. Hit us up on the socials, Dan and I spend way too much on there, primarily on X, formerly known as Twitter, and LinkedIn. Thanks for tuning in. Bye-bye.

Author Information

Daniel is the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise.

From the leading edge of AI to global technology policy, Daniel makes the connections between business, people and tech that are required for companies to benefit most from their technology investments. Daniel is a top 5 globally ranked industry analyst and his ideas are regularly cited or shared in television appearances by CNBC, Bloomberg, Wall Street Journal and hundreds of other sites around the world.

A 7x Best-Selling Author including his most recent book “Human/Machine.” Daniel is also a Forbes and MarketWatch (Dow Jones) contributor.

An MBA and Former Graduate Adjunct Faculty, Daniel is an Austin Texas transplant after 40 years in Chicago. His speaking takes him around the world each year as he shares his vision of the role technology will play in our future.

SHARE:

Latest Insights:

Entry, Mid-Level Models Plus More on Cyber Resiliency
Camberley Bates at The Futurum Group covers NetApp’s announcement on block, cyber resiliency, and BlueXP.
Zoho Unveiled a New Agentic Platform to Deliver AI-enhanced Portfolio-wide Improvements Including Zoho IoT
Futurum’s Ron Westfall explores how Zoho is positioned to develop AI-enhanced solutions that can drive customer value including the optimization and full security of IoT assets that improve business outcomes and overall experience.
Microsoft Adopts a Paid Metered Approach for Autonomous Agents, Highlighting Evolving AI Monetization in Microsoft 365 Copilot
Keith Kirkpatrick, Research Director at The Futurum Group, examines Microsoft’s shift to metered pricing for autonomous agents in Microsoft 365 Copilot and its impact on enterprise AI adoption.
Andi Gutmans, VP & GM of Databases at Google Cloud, and Karan Batta, SVP of Oracle Cloud Infrastructure, join Patrick Moorhead to share their insights on leveraging the best of Oracle Database and Google Cloud for organizational growth. This collaboration aims to provide enhanced data management solutions.

Thank you, we received your request, a member of our team will be in contact with you.