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:
- Cisco Gets AI Lift, All-In On DPUs
- DOGE & Tech Impact
- Lattice Semi Q4FY24 Earnings
- Elon Makes A Play For OpenAI
- CAPEX Spending Still On A Tear (Told Ya’)
- Intel Rising, But Why?
- JD Vance Lays Out US AI Case In Paris; Will EU Ever Change?
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, and it is the Arctic temperatures here from Austin, Texas, hence, the fur. I had to break out the Aspen clothes though, Daniel. You know, why not?
Daniel Newman: Yeah. You saw me yesterday, I was on a call. I think I had a hoodie sweatshirt on and a winter coat wearing two hoods and a hat. The thin skin, it doesn’t take long. Having been from a cold climate over two to three years, I’ve completely lost my will to leave and go anywhere that’s cold.
Patrick Moorhead: I mean if you have a choice, why do it? What I do is look at all the winter clothes that I have in there just sitting, and I feel like I’ve got to use those pretty quickly. But anyways, it is episode 249. By the way, for our Valentine’s Day dinner tonight, make sure you bring your puffy, okay?
Daniel Newman: I am all over it, man. I’m so glad that we shared that with everybody. Before we do our real Valentine’s dinner, we’re going to do our real Valentine’s Day dinner.
Patrick Moorhead: That’s right. Don’t tell anybody, though.
Daniel Newman: Yeah, you and me. Oh my God. Gosh, why do we do that?
Patrick Moorhead: Anyways, folks, it’s episode 249. We did make it without getting canceled. I do feel like though, the next four years the chances of getting canceled have decreased a lot, but we’ve got an amazing show for you. We’re going to be talking about Cisco earnings. Did they get an AI lift? What’s going on with all this DPU conversation? We’ve got DOGE. We’re going to talk about DOGE and the potential tech impact, definitely not as straightforward as you might think. Lattice had Q4 earnings and also 2024 earnings, we’re going to break that down. Did they hit a bottom for industrial? We will see. Elon’s making a play for AI, OpenAI or is he? We’re going to be talking about CapEx spending with the Mag Seven, or Daniel, should I say the Mag 1.5? Maybe we can have a little fun with that as well. Intel stock is absolutely on a rocket ship but why? Probably like five different scenarios out there that we can bounce back and forth. And finally, JD Vance heads to Paris to give the US case for AI, but will the EU ever change? Anyways, Daniel, let’s kick this show off. Control room didn’t advance to CHIRON. Shame on you, guys. Get your act together. Anyways, let’s get into Cisco here, let’s break down the earnings. They also coincidentally had their big EMEA show, their big tent event over there, boy, during the quiet period.
Daniel Newman: Yeah. Well, sometimes when you’re talking about tech, it’s not about stocks. So here on the Fintwit side of the world, where you and I are constantly forced to swing the pendulum back between talking about what these companies are doing technically and then always back to markets, that’s what moves things, of course. But look, Cisco is doing the deed. I say every quarter when companies like Cisco and HPE, and IBM and some of these big B2B enterprise companies show up. I say, “These are bellwethers. These are the bellwethers for the enterprise.” We love talking about the Mag Seven, the big hyperscalers. We love talking about the $70, $80, $100 billion CapEx spends. But the average enterprise in the CIOs and the ITDMs, and how they’re continuing to think about and invest and build technology that serves these companies today.
So Cisco has been one of these companies that’s been a bit of a meme, and the meme has been somewhat cynical that NVIDIA is the Cisco of the AI. You’ll see they show the curve of the rise of the stocks over a period of time, and then the fall off that took place when the dot bomb happened with Cisco having been one of the big winners. I continue to reject that notion, and continue to actually believe that there’s a couple of different things happening, but Cisco is performing. Cisco beat, Cisco raised. And one of the things, Pat, that I come out every quarter and I just beat this drum of, “Hey, you want to win over the minds and hearts of the investor class right now, you’ve got to talk about how AI is shaping your business.” And finally, we got some details from Cisco CEO Chuck Robbins, and CFO Scott Herron about the business about AI.
Well, in their business that’s now 56% software, I think it’s 56%. They also announced their AI infrastructure orders, which to date, is still a smaller number in the grand scheme of the size of Cisco. At a company that’s pushing north of $50 billion of revenue, but $350 million, Pat, in AI infrastructure orders this quarter. They had $700 million in the first half, and they’re expecting now over a billion of orders for the fiscal year for their web scale. And remember, their fiscal year runs mid-year to mid-year, so that means you’re accelerating into what will be a new fiscal year. They did see big product order growth in the 29%, 11% if you don’t count Splunk. And they were also very able to do a bit of a victory lap at how successful and quickly Splunk has been integrated into the business and its growth and its strength. And that certainly added a big software mix to the company, it added a big diversification to the company. And they’ve been able to take the EPS and make it a creative earlier than they had originally anticipated.
They’re seeing big growth in ARR, big growth in software. They had the big AI defense. If you didn’t check it out, we sat down with both Chuck Robbins and Jeetu Patel at Davos ahead of this, and they had a lot to be proud of. But across their portfolio, very diversified into that security network extending more and more into compute and now software, Pat. And of course, the company continues to return equity, return money to shareholders. They returned $2.8 billion this quarter, so good strength, good diversification. The AI story is starting to come to fruition. Splunk integration is happening. This AI defense offering is certainly going to be important as enterprises start to look at deploying AI at scale. And I like what they’re doing and I like how it represents a broadening out that I keep talking about of AI, beyond just a couple of infrastructure and cloud players.
Patrick Moorhead: Yeah. That’s a good breakdown, Dan, and as you said, a beat, a beat and a raise. And the revenue beat was in line with the previous three quarters, and if I average out the EPS beat, it was also averaged out. I had expected the stock to fly even bigger, but there was a lot of discussion about orders’ deceleration. And I think Cisco had a good response to that attributed to the longer AI deal length. These deals are a lot, are very, very complex. The other thing is, is if you look at the last six months of Cisco stock, it’s up 41%. It’s a real screamer. And by the way, if you compare that to the Mag, the Mag Seven, NVIDIA, AMD and folks like that, it’s very, very good with that comparison.
So a couple of things. First of all, I appreciated, I had a conversation with Scott Herren after the earnings call. Gave some color and some real insights into a few things, but some of the highlights that I saw. First off, one of the challenges that Cisco had previous two quarters was, “Hey, how are you taking advantage of AI?” And what I really like is, again, there’s only so much real estate you can put into an earnings deck. But I still have to give them credit for the way that they laid that out, which is AI training infrastructure, AI inference, and AI connectivity. And unsurprisingly, you saw their stock bump up since then. I’m not saying, “Oh, because you added this slide,” but I think what investors want is detail. Give me the thesis and then show me the numbers, show me some proof of life that you have something going on here. Another thing that I appreciated, which was a carryover from EMEA Cisco Live, was AMD DPU integration and Hypershield DPU integration to the 9,300 switch and Hypershield integration.
So DPU is essentially intelligent networking at the edge. Hyperscalers have had this for years. We talk about AWS Nitro, it’s very similar to that. This is the version for enterprises. And DPU providers, Marvell has one, AMD has one, and NVIDIA has one, so I see this as an untapped opportunity for the company. There was some discussion on lower gross margins, but guess what? This is supply chain shifts to get ahead of an expected tariff situation. This is the one thing I really appreciated from my conversation with Scott, as Scott and I went deep on different countries, different where they do manufacturing, where they have puts and takes. How they’re playing this chess game and, “We’re 80% mitigated.”
So best conversation I’ve had during this earnings season and probably last earnings season related to that. A couple other announcements. Gary Steele moving on, he’s going to be CEO of a noncompeting company. Well, actually I’m not surprised at all that he’s leaving, but I thought this Cisco is going to be his coup de grace, like his swan song, but nope. Gary’s a very successful, competent CEO guy and he’s going to do the CEO thing again, so best of luck. Don’t know where he’s going yet, but I’m going to be checking that out.
Daniel Newman: You could have seen that one a million miles away. Gary’s the come in and turn the company, sell it, take it public guy. When Splunk brought him in, it was all about the transition from Doug Merritt to him was about getting a deal done. So he did it at Proofpoint, he’s doing it here. He’ll do it at the next thing.
Patrick Moorhead: So what do you think, startup or big cap or small cap?
Daniel Newman: I would say it’d be more of a small mid-cap, meaning I think it’ll probably be a publicly traded company. It just seems to be a bit of his MO where he’s either going to be taking it back private, getting it sold. Or potentially like you said, a company that’s close to going IPO and being the kind of guy to get that done.
Patrick Moorhead: Yeah. Final comment, I’m looking forward to learning more about Silicon One and what they’re doing. You might’ve noticed, they’re starting to ratchet up the disclosures on that. I thought they should have done this a year ago, but I think it’s in a much better place than it is right now. So let’s move on to the next.
So DOGE, which is Department of Government Efficiency, which is led by Elon Musk himself, has brought in some very, very smart and reportedly very young programmers and data jockeys to come in and identify where the money is being spent, who’s spending it, and what they’re spending it on. I would say it’s probably 25% of the conversation out there for the new administration. And by the way, trying to get the government more efficient, there’s nothing new about that. We saw Obama, we saw Bill Clinton, and then we saw presidents like McKinley and folks like that. If you check out history, they had various levels of success. Reagan talked a good game, a big game. He didn’t actually make a lot of cuts, but what he did is he made a lot of growth, okay? I can imagine during stagflation, you don’t want to cut, you had to bring down inflation because it was like 13%, but anyways, I digress.
So what is the impact on tech? So first and foremost, it’s not clear. You’ve got a couple scenarios going on right now, and none of this can be confused with the potential impacts of tariffs, which could have an impact. But the way that I look at that is if I look at the I’m going to take a bull and a bear case, okay? So on a bull case, you have a government that has aged systems, okay? And whether it’s TSA, whether it’s the IRS, whether it’s all these government agencies that may or may not get deleted, and that’s the bull case to modernize. Whether it’s private cloud, a little bit of public cloud, but there’s a lot of software and applications that need to be modernized, and that would be the bull case because you can’t do this. It’s very hard to do this without upgrading all of your infrastructure to something that’s modern and something that can be cloud-enabled.
The bear case is very simple. The budget got whacked by 50%, and with that gets a reduction in that agency’s IT budget by 50%. And I guess the third scenario could be it’s a mix of both, where you cut an agency’s budget, overall budget, it’s primarily people. And you cut some of the programs that aren’t deemed necessary or have a decent payback. And then you take some of that money and then you invest in infrastructure. So Daniel, where I am here is more in scenario, not laying it out there. I have not heard enough to take the shot on this one.
Daniel Newman: Yeah. Look, tech is incredibly deflationary, printing money is inflationary. That really is, by definition, inflation is printing money, and a lot of people don’t understand that. But running a trillion-dollar deficit, we are spending more servicing our debt annually than we spend on our defense budget at this point. And our defense budget is full of waste, as is our every other budget. Look, the biggest thing is I can understand people not liking Elon Musk. I can understand that. I can understand people saying, “We didn’t elect Elon Musk.” I would also say we didn’t elect Fauci, we didn’t elect Bill Gates. We didn’t elect any number of bureaucrats and influencers of bureaucracy over the last few decades, but at this particular moment, he’s the person in focus. Having said that, you run companies, Pat, even in your own household. Look, when you have seven subscriptions to Netflix for two people and you’re paying them.
At some point you look at that and say, “We shouldn’t do that. It’s just waste.”
So first and foremost, the government has a fiduciary responsibility to us, the citizens, the taxpayers, the people that actually write the checks to these people, to try to spend the money with a level of fiduciary responsibility, with a maturity that we just aren’t seeing. So before we even get to the, “Should we do DEI? Should we do ESG?” What I’m just saying is there is so much irresponsibility that’s showing up immediately with some of the ways we spend money, that I would think every person on the planet would say, “Thank you, Elon. Thank you, DOGE, for actually…” And I know I’m going down a bit of a tear here, but look, I am impassioned right now by what I’m seeing happen. I believe that we can spend better. I look at, this is like I said, a little bit of a diatribe off tech.
But look, I’m going to take the path, we want to raise our children to be productive citizens in the world in the future, I want my dollars to go to my schools. I want my kids’ curriculum to focus on giving them the kind of education, including on technology, on AI, on quantum mechanics, physics, chemistry, that is going to make them able to be successful and productive in this world. And when I see just the absolute irresponsibility and waste that we spend. And not to mention the money we send overseas while we continue to neglect our children, our veterans and everything here.
Pat, but overall, we can’t actually cut enough spend single-handedly, to actually beat what we have going on in inflation right now. We also can’t make the government the only and the biggest employer in this country. All these projects that we’re funding that aren’t actually being executed, infrastructure projects, Pat. The ones like, “Hey, we’re going to put EV chargers.” Why not let the private sector run with this? Why not incentivize private companies to build solutions for these things? I see it with AI, I see it with EVs, I see it with robotics. I just genuinely believe that technology is deflationary, cutting waste is deflationary. Reallocating dollars towards things that actually do build the infrastructure and the capacity, like the CHIPS thing, Pat. We talk about $50 billion for CHIPS Act like it’s some huge numbers.
Patrick Moorhead: Yeah.
Daniel Newman: We’re sending $50 billion overseas. What did I hear that like Zelenskyy couldn’t account for $70 of the $166 billion that we’d supposedly sent to Ukraine? There’s more money that is unaccounted for that was sent over for a war that we really don’t know what is happening. Then we plan to spend in entirety to be the world’s leader in manufacturing the most important semiconductors for not just tech, for actually the future of economic dominance in the region.
Patrick Moorhead: Yeah.
Daniel Newman: I want to say more than this, we need to keep going down this path. We need to keep cutting and give our children a chance to be able to foot the bill in the future.
I like what’s going on, I really do. I’m really pleased with it all.
Patrick Moorhead: Man, great, great breakdown. You gave our listeners more than they probably thought they were going to get and I appreciate it. And Daniel, I know you’re a young spry and I will never hold your lack of experience and age against you.
Daniel Newman: Thank you, thank you.
Patrick Moorhead: But there was definitely a time where pretty much everybody was on, it was an affront to society to run a deficit. It really was, and somehow we lost the recipe. And by the way, this isn’t an R&D thing, like a Republican and Democrat thing. The entire bureaucracy of the government lost the plot. Voters didn’t call them on it. And if the people who were in power wanted to stay in power, they would get fatty programs for their constituents. So yeah, anyway.
Daniel Newman: There’s a difference, by the way, between waste and fraud. The word fraud is being thrown around a lot, and some of it may deterministically become that. But I just don’t think across the board we can argue there’s been waste. And by the way, waste to your point, it’s not a partisan thing. The waste is happening across the continuum. In the end of the exercise, it’s the same thing, man. If you came home and you found out you had 17 subscriptions to the same thing, you would cancel 16 of them.
Patrick Moorhead: Right.
Daniel Newman: We as a government, and we as the people that pay the bill, should just expect the people that we’re sending money to, to spend it like it’s their own. And that’s what I feel like a lot of people aren’t upset enough about this, is like when you’re taking 40% and 50% of your income per year. Have you ever seen Charlie Kirk do the whole breakdown of how your dollar is taxed between when you make it and when you spend it, how many times we’re taxed? It’s an incredible little clip, I’ll have to share it with you. But the point is, is by the time we actually make a dollar and pay 50 cents, Pat, and then we go back out and spend that dollar and pay tax on the dollar on the way out. And then you pay on the property and you continuously pay on your property after that. We are taxed well over 100% on every dollar we make, and that’s super inflationary, so it’s crazy.
Patrick Moorhead: Yeah. One of the memes I really like is we had the Boston Tea Party getting taxed at 1% to 2%, and we’re all slumbering at our 39% tax rate and not getting a whole lot for that. Okay. Hey, let’s move on. We could probably stew on this for an eternity, but Lattice Semiconductor, new CEO, second earnings call. How’d they do this quarter, Daniel?
Daniel Newman: Yeah. This one, we go from big, super macro to a very narrow subset of the semiconductor space, Pat. But look, you and I spent some time with Ford Tamer, their new CEO, we’ve gotten to know very well their leadership. Esam Elashmawi, their chief strategy officer, spent some time with him on earnings day. We’ve spent time over the last few years, as they’ve built out the developer ecosystem. Important company that built some really critical security technology, some AI features that go on to things like PCs and servers that people don’t talk a lot about. But this company’s also been a bit of a litmus test for the recovery of the broader semiconductor space. For everybody out there that looks at it like, “The semi market is great because TSMC is doing well and NVIDIA is doing well,” is missing the point that everything outside of AI chips has been really slow to recover. Even PCs and handsets have not recovered nearly as quick, long-time inventory turn. The AI PC hasn’t necessarily driven that huge cycle that people expected. Communications, networks, industrial have all been slow to recover. The good news was in the conversation that I had with Esam heading into this one, is that the company believes that it’s at the endpoint of its U-shaped recovery. So it wasn’t a V here where it was for others. It’s been much more of a U for these companies that have high dependence on consumer, commercial, communications.
So they think they’re coming to the end. They’re actually booking is accelerating to be faster than their invoicing, which is always really important to get that number above one to show that you’re entering a growth era again. They’re seeing double-digit growth in their mid-tier products, which has been a big bet for that company. They were big, they were always at that very low end, and that’s where they differentiated. With AMD acquiring Xilinx, with Altera being spun off, the upmarket part of the FPGA space became an opportunity. AMD is very focused on leveraging that technology and what it’s building at scale, but hasn’t been as focused narrowly on Xilinx. Altera is still trying to find its footing amidst the Intel change, which we’ll talk more about Intel later, actually had a good week for once. But then Lattice made this big bet with Avant going upmarket, building a more competitive, mid-tier product. And it’s seeing double-digit growth there so that’s really good.
The other thing is they actually talked about AI impacting revenue. They were talking about a double-digit, about 17% was the number they threw out. Need to dig a little bit more into how they’re calculating that exactly, because it’s like we want those AI readouts, but exactly what do they mean by that? I’m guessing that it’s where the sensor, the FPGA is being used for some sort of AI-specific use case, workload that they can point to. And then finally, Pat, I just made one other comment was we had Jim Anderson, their former CEO, from Coherent on our show. He left, their CFO left when Sherri Luther left just after he left. I think she went with him, right? She’s over at Coherent now with him.
Patrick Moorhead: Yes.
Daniel Newman: Which happens a lot, but those have been big holes. Esam took the interim CEO role, they brought in Ford. And they had some holes in the leadership team and they were able to announce a bunch of new leaders that were brought in to fill gaps. So I think they got a lot of the right momentum, Pat. This one’s another one of those test case data points for the overall shape of the semiconductor market moving in the right direction, and been able to weather a storm, keep margins intact. And I think this market will continue to accelerate into future quarters.
Patrick Moorhead: That’s a great breakdown, Daniel. So I had a call as well going through this, which I always, always appreciate. And I’m going to try to fill in the cracks here, but let me give you some macro view here. The stock is up last five days 19%, and the last six months it’s up 36%. Now, this stock has always been a rocket ship, but what happened is the market pretty much just cratered, particularly for industrial type of applications. Computing is on fire, but industrial and even telco and things like that are in the toilet. And as you guys know, I always try to discern between, “Hey, what is a company-inflicted wound and a market-inflicted wound?” Well, the bottom has been hit from a market point of view, and I believe that we’re back in growth mode. Now, everybody’s being very cautious, including Lattice on this, but at least the numbers if you look at quarter-on-quarter, you look at year-on-year, there is a sign of a market recovery. And on the call, they talked about a book-to-bill ratio above one for the first time in six quarters. And the demand is even outpacing, at this point, shipments.
So that is a huge sign and that now hopefully, we can get back to some normalcy, where it’s looking at competitiveness versus Altera and versus AMD. And what type of impact is Avant making on the entire business? And just so y’all know, Avant is the mid-range, higher performance, still low power, but higher power than its predecessor that’s in the market as well. And this is an absolute new market for the company. So even missing EPS expectations and a little pressure on gross margin, that’s really attributed to some packaging write-downs. The reaction to the earnings was super positive, up 18% in the last five days. So this is good to see a very well-run company moving forward.
I’m looking forward to, sorry, their Financial Industry Analyst Day, Equities Analyst Day in the future here to better understand the strategic moves that Ford is going to make. He’s definitely an M&A guy, he’s an operator. He’s got so much experience, but I don’t think Ford came in to sit on the stock and get 5% gains every year or even organic growth. I do think we’re going to see some acquisitions here moving forward, so I don’t say this for everybody, but I think the best is yet to come. All right, let’s dive into another topic. And by the way, probably by the end of this month, I know we keep teasing our new format. We’re probably going to be digging into topics like this a little bit more, as opposed to getting into specific company and specific company announcements.
But hey, with that said, Elon is making a play for OpenAI or is he? A little history here. Elon was the first and largest investor into OpenAI, which was the non-for-profit company that was going to build models to empower the planet. I think this was in their, I don’t know, mission statement, not the exact words, but you get it. And then OpenAI gets in and starts charging a bunch of money with their services. They partner with Microsoft and they’re generating an incredible amount of revenue and losing an incredible amount of money. Elon had bailed at some point, at which I think he deemed that they weren’t going to be successful or weren’t interesting. But then when OpenAI started charging a bunch of money and not doing what, let’s say, Meta’s doing with Llama, things got a little dicey. There were lawsuits that were filed, Elon filed them against OpenAI. There was a social media attack, I think he called it Closed AI. Even suggested and Elon had-
Daniel Newman: Scam Altman.
Patrick Moorhead: Yeah, Scam Altman. He had a temper tantrum and a meltdown here. So there have been talks about separation between the nonprofit OpenAI, and the for-profit AI.
Investors want to see some paybacks. Sam Altman, who does not officially have any ownership stake in OpenAI, he would like some payment too. By the way, Sam has really nice cars, by the way.
Daniel Newman: Yes. For a guy that owns none of OpenAI and does this all out of the goodness of his heart, that Koenigsegg is a nice, I think, that’s like a $3.5 million car.
Patrick Moorhead: Yeah, but I digress. So essentially, when you’re going for another round of funding and you’re trying to separate from the nonprofit mothership, it all comes down to valuation. And there are some questions on whether this, what was it, a $90, $95 or so billion? I don’t know the exact number here, but essentially the debate is, is this real or is this to slow down and make this process of valuation and divestiture from the mothership more rocky than it is? So anyways, I do believe this is just yet another way to interfere with this. I do understand where Elon’s coming from. He pours all this money in under the guise and then he sees no value. I also do think even though Elon hasn’t talked about it, is he’s petrified about unbridled AGI and what it can do to the planet. So Daniel, I’ll pass the baton over to you.
Daniel Newman: Yeah. If I was smart enough to be inside his head, I would probably have a better opinion, but let’s just say I have a few things. I think one is there’s some gamesmanship. I think when you function, live and operate at the level that Elon and Sam Altman live at, that there is some, the combination of utilizing the court of public opinion, the utilization of the actual courts, the leverage, some of this is also about staking and creating some tension in valuation. Right now, they’re talking about raising another $20 or so billion, maybe more from SoftBank at a $240 billion valuation. Absurd, you’re talking about a company that’s losing billions of dollars every year and burning cash. They have some important IP, but very interesting how you come to that valuation. There’s an increased amount of competition in this space. Just a few weeks ago, we thought DeepSeek was the end of OpenAI, and now you’re going to tell me they’re worth another $100 billion more than they were worth three months ago?
Now you put some question marks around the real valuation. You put some consternation in the board. You create some division potentially between the board and the CEO, which there’s already been in this company in the past. Certainly, as you said, Pat, Elon has actually publicly said he’s uncomfortable with what AGI means to the future. He seems to have a different opinion about it than maybe Sam Altman does. At the same time also, the idea that OpenAI completely abandoned its mission, it abandoned its early investors. It abandoned its commitments to being open and open-source, and open research. When you and I met with Yann LeCun, you couldn’t hear the people in these statures that really are the leading thinkers in the world about what AI is and should be and what the process is. And going to AGI, nobody that really seems to understand it, seems to think it’s a great idea for this to be a black box. So I think there’s all kinds of different games being played here, and I think in the end, Elon wants to slow it down. And if there’s anyone in the world that can do it, it’s him.
Patrick Moorhead: Totally. Yeah, I’m looking forward to this, just if nothing, getting the popcorn out, the entertainment. So Dan, let’s move to this next one. DeepSeek was going to be the end of CapEx, the end of NVIDIA. And we were going to be running all AI on the smallest GPUs that were 10 years old, because DeepSeek pulled out a miracle, invested $5 million in this model. What’s going on with CapEx at the Mag Seven?
Daniel Newman: They invested 5 million bucks, and came up with better products and solutions than hundreds of billions of CapEx.
Patrick Moorhead: Totally unbelievable.
Daniel Newman: Now we’ve had six of the Mag Seven report, and the biggest of the big hyperscalers have all come out, Meta, Amazon, Google, Microsoft, all north of $65 billion of intended CapEx spend. And guess what they’re going to spend most of it on? GPUs and AI infrastructure. So while all of us were spinning our heads around and doing media hits, and talking endlessly about the world ending and DeepSeek being the eradication of all things AI, and that the bubble was about to burst. Every major CEO of major Mag Seven company right now is basically saying, “We’re full throttle on spending.” Even as they continue to report, and you heard our shows over the last few weeks, that the growth and the ability to articulate the consumption number is not actually proportionally growing nearly as fast as the CapEx is growing. Why? Because we genuinely still believe that we’re very early in this workload. And I think when you hear numbers of inference being 20 times, I’m seeing things that could say it’s thousands of times in orders of magnitude larger workload.
So in order to support that, the investment has to be made. So whether it’s been Meta, Microsoft, Amazon, Google, all of them are going to continue to spend. And Pat, there was some interesting research you shared with me and it’s been kicking around about what this means in GPUs and XPUs. I think there’s just this big problem in the market right now of people not being able to think about two things at the same time. So the two considerations at the same time that are going on, is one is there are going to be aggressive work. There is going to be aggressive work done by research, by technology companies, by labs, by software to figure out how to more efficiently deliver AI. I shared a post the other day with IBM about Granite and how they’re delivering 4% to 6% better performance on specific workload using Granite and InstructLab against GPT and Llama.
And doing it at 75% to 96% or 94% lower cost.
This is where we want to go, but the point isn’t that once we do that, it’s like we’re going to keep and use the same amount of compute. The point is, is once we figure out how to do that, the compute is going to be exponential by many orders of magnitude. So the idea is yes, more efficient models. Yes, a narrower, smaller language containers, specialized services software, but we’re going to do both. And then on top of that, Pat, I think we’re going to augment that the NVIDIA leading edge, most performant, lowest number of GPUs to do a certain type of training or inference workload, will continue to be rolled out for that purpose of advancing the AI future. At the same time, companies like Meta, you’re hearing more about them building their own chips. Google’s doing it with TPU and is doing it with arm off the shelf. Amazon’s been doing it for years with the Trainium and now Trainium 2, and Microsoft as well. They’re going to continue to build ASICs.
And we put out some research through Signal about Intel Gaudi, they’re going to put out an ASIC, and they’re going to eventually build a GPU. AMD is going to continue to invest and build Instinct, because we’re going to need more compute. The weird zero-sum thing that’s happening, and I’m going to end here. The zero-sum thing that’s happening, the idea is that as we get more efficient in being able to deliver AI, that we are going to have some finite utilization of compute is the most… It’s embarrassingly misunderstood by people who should know better. And as they know better, it’s like we have barely even started to scratch the surface for how we will utilize AI. So the need for more efficient compute and different architectures and scale, it’s not the end of NVIDIA, people. But it also means there is a broadening out and a rounding out and opportunities for a number of other companies to participate, to support a much bigger set of workloads. And CapEx will keep roaring and the AI trade is far from over.
Patrick Moorhead: I don’t even know if I’ve got anything to add, other than I’m going to do a victory lap. It was good. I’m just going to do a victory lap, which was when DeepSeek hit. Well, actually I’m going to do an extended victory lap. Every time that I get on broadcast, anytime I get asked, “when does this gravy train slow?” This was pre-DeepSeek, I said, “Essentially, when we hit AGI, and that inference is on every single endpoint on the planet, okay, and on the edge. And until we get there, this gravy train is going to go.” Then a more pragmatic version of that, how long will the NVIDIA train go? It’s at least a two-year run and I feel pretty comfortable in that. Maybe it’s 18 months now when I started talking about that, but when DeepSeek hit, I had the same. A couple of people don’t understand is that this new version of AI doesn’t require… Well, it requires more than these very narrow L40s that NVIDIA put out. Latency is paramount and you just can’t get that latency with a trimmed-down GPU. So yeah, I don’t see anything, I don’t see this train slowing until we hit AGI everywhere. Training, pre-training, inference in phones, in PCs, on the industrial edge and there we go. It’s tough winning all the time, Dan. That everybody just expects you to just call it right every single time, which by the way-
Daniel Newman: I’m getting tired of being right. It’s exhausting.
Patrick Moorhead: Yeah. I do want to share though, out of the other side of mouth, I did have a major miss. I did think at the second half of 2024, we would see a big spike in PCs and smartphones. I made two very big mistakes. The first one was I didn’t expect Microsoft to slip its schedule for its premier AI feature in Windows by nine to 10 months. Okay? It was supposed to be out in May, it’s still not out. It’s freaking February and that thing is sitting in the Windows insider. I think it’s great. And the second thing I didn’t consider was just how horrible Apple would be on their AI and where they are now. It’s a freaking laughing stock. The big advertising is on emojis, which is just completely ridiculous. I do think that every once in a while, their email summaries are interesting, but maybe it has a 50% hit rate, the summary of the notifications. And even though Samsung, that has a vastly improved and superior phone AI, doesn’t have the market impact of what Apple can bring to get end consumers excited about it. And by the way, if you’ve got a new S25, man, that generative fill is unbelievable. Try it out. Anyways, I don’t know how I got on generative fill in S25 from a CapEx discussion, but here we are. Okay.
Daniel Newman: I’m proud of you.
Patrick Moorhead: You got it, buddy. So for the show and the discussion, I think many of you are in for, thank you, we have almost 300 people live on X watching. Intel is seeing its best week since 1975. Daniel, your parents were probably in college then or very young. You hadn’t even been thought of for a long time. I was seven years old in 1975, but what on earth is going on here? I got an hour’s notice to get on CNBC’s Fast Money to lay this thing out, so I was literally just going off the fly here, but let me break this down. So there’s two things going on. I would say primarily, it’s about the rumors of some sort of a tie-up with TSMC, and I’ll go through some of the potential scenarios on that. JD Vance made some comments that definitely could be interpreted as Intel-friendly.
And then I would say the third thing is we’re starting to see some very positive leaks about 18A out there, but let’s talk about TSMC, because I think that was the first driver. TSMC had their board of directors meeting here in the United States, and there was some cross-correlation between that, but there’s a lot going on here. What I said yesterday, and it’s funny, I hadn’t even seen all the rumors that were out there on TSMC, is that an outright acquisition is off the table. It would be a national security nightmare. And oh, by the way, getting antitrust approval, and remember, it’s not just antitrust approval in the United States. It’s antitrust approval globally that it becomes an issue.
Now, what I said on Fast Money last night, is I could see a JV happening, but there would need to be some intellectual property that goes into it potentially. And my gosh, the competitors and the crown jewels from a TSMC point of view, is there intellectual property on how to move that forward? I also don’t think that Intel needs help from TSMC on big die. I think 18A is from, again, a big die perspective is looking really good in performance, good in power. Yields probably aren’t as good as TSMC at this standpoint, but yields at this point are reasonable, but there’s a lot of work to go. I also believe that the Trump administration is putting in parallel, an immense amount of pressure on NVIDIA, Broadcom, AMD, and Marvell to do a double-take on 18A.
There’s also been floated a 15% tax break on technology manufactured in the United States and a potential tariff, big tariffs that could be levied against Taiwan, so there’s a lot of moving pieces here. By the way, on the tariffs I’ve seen, “Oh, there’s no packaging done inside of the US, so TSMC won’t be impacted by the tariffs.” Come on, folks. The price will be pushed to somebody and the end price will go up, so don’t even think for a second this could be the case. I did see an interesting rumor out there about a consortium. And I’ve talked about a consortium before, which says you have AMD’s, Marvell’s, NVIDIA’s, Broadcom, Apple, Qualcomm belly up to the bar and put some investment in here, in addition to TSMC. But aside from cash, do you think TSMC wants to put in IP and is weighing this IP versus tariffs? Anyways, it’s not clean-cut here, but anyways, that’s what I think is going on with Intel at this point.
Daniel Newman: So it is taking a bit of a hit today. It’s down almost a buck, as we speak, after what’s been a more than 20% run. I saw someone show year-to-date, Intel was up more than NVIDIA and I think that’s funny that people are doing that. But I think, Pat, you’re onto a lot of what I would consider to be the important points, one, trading near book value. Basically, when you get to that point, it basically means the company’s been left for dead.
Patrick Moorhead: Yeah.
Daniel Newman: It means that basically people think that the stock is worth the net sum of its assets. And when you have 65%, 70% of the PC market still, you still have a big share of the data center CPU market. You have a massive investment in a foundry business, in an era where building leading-edge capacity in the US is a hot button item. There has to be like a, “Hey, if I’m going to make a bet here of how much upside there might be, there is a lot of upside sitting in Intel.” I also think people are appreciative with MJ and her sort of, I think she’s narrowing the scope and I think people wanted that. Intel was in a lot of things, they’ve divested, they’ve spun off. She really seems to focus on protecting market share in the PC and workstation space. I actually think they got a boost from canceling the Falcon Shores. I think they actually, because people just did not believe it was going to be competitive. And I think to some extent like, “Oh my gosh, they actually are recognizing that they can’t keep serving up inferior products and expecting people to buy them.”
I think people want them to have a competitive data center part for AI sooner. We’ve tested in Signal65, we’ve tested Gaudi 3. As an inference part, it’s pretty competitive. It’s not going to meet the software requirements to be competitive for any sort of training at scale. But if you’re using it as an inference workload in enterprise, I think they’ve got a chance to do more with that. Is it going to be a $50 billion or even a $10 billion or even a $5? Probably not, but could it be a billion-dollar revenue stream for them next year? Yes. Could they learn something from that and the test chip with Falcon Shores? Yes. I’ll leave it at this. Trump’s legacy is about making sure the US has shored up the leading-edge manufacturing. We are very closely aligned to Taiwan, but Taiwan’s a risk and he knows it, and he knows it’s a risk. I think the negotiation here is to get Taiwan to bring its most advanced technology to the US. It’s to maybe potentially share some of the critical IP. People also have to realize it’s possible we’re going to have to take a sidestep or a step even somewhat backwards to go forward. Meaning that all the current, fabulous designers are more than happy to just use TSMC. They are, they’ve all told me. But having said that, policy is going to be what ultimately determines whether or not more leading edge. You talked about in your post about some good content, some of the big die on 18A. Some of the smaller die opportunities in 14A, bringing maybe potentially mobile parts.
But to some extent, if the government basically says, “You have to do this,” or they tariff in tax and make it so unlucrative to use TSMC. Or through this JV thing, make it where it’s a forcing function for TSMC, Intel to work together to make it continuously be lucrative to all the fabulous, I could see that happening. Right now, there’s just not enough incentive. I talked to Bloomberg yesterday. They asked me, they go, “What’s the difference between the Biden CHIPS Act and what Trump’s thinking and how he wants to renegotiate, is that there was no teeth in the Biden CHIPS Act.” There was all the money being given, but there was no forcing function to get people to manufacture more here. So I’ll leave it at that. I think there’s a lot of upside in Intel, not a lot more downside.
Patrick Moorhead: Let’s go to our final topic here. JD Vance went to the Paris AI conference, laid out a few things for AI. EU had a big game, Chrome talking big here. Dan, let’s balance these things against each other.
Daniel Newman: Yeah. Firstly, whether you like JD Vance’s politics, I like him as an orator. He’s a good speaker. He’s very articulate. I’ve been saying the same thing, making the rounds in media about US exceptionalism, about US wanting to maintain its leading position as the world’s leader in AI. When DeepSeek happened, that was the first thing I was screaming at the top of my lungs that, “Hey, we need to continue to invest to make sure we never risk not being the leader.” We’ve made the jokes on this show and elsewhere, Pat, about the EU basically regulating itself out of every major innovation cycle over the last, I don’t know, 100 years. Europe had a lot of its own exceptionalism for a long time, but now it feels more like tourism and food and not enough innovation, and we want Europe to be a more innovation-driven partner. It’s been more of a regulation, policy and taxation obstacle.
And of course, what I do like that comes out of Europe at times, is they do focus on things like personal privacy, data security at a level where maybe the US in its search for global leadership, tends to abandon security and privacy first. But I think his call, Pat, was exactly this, less regulation. Stop tightening screws on things and setting yourselves behind the curve. How can you work together? There has to be some level of ideological alignment that’s going to take place between the US and its pace of innovation in Europe and its regulation in order to get there. And then, of course, he subtly even talked to the fact that Europe has been a little bit more inviting to what I would consider to be authoritarian and conflicting government regimes. There seems to be more interest in alignment. You see it through the shifting in population across Europe.
And I think he was pretty stern in his warning against basically, where do you choose to partner in Europe? Are you going to choose to partner with the US and the west, or are you going to choose to partner elsewhere? So I think he came down, he was very clear, he was very articulate. But he was very much pushing that we will maintain the leadership here in the US, and this is the terms and conditions for you to come along.
Patrick Moorhead: Excellent. Listen, yeah, JD Vance just he knocked it out of the park there, and his big thing was this conference is about growth, not about, what word did he use? AI safety. He said it’s about AI opportunity versus AI safety. He was very clear on, “Stop messing with US, AI companies.” I felt like it was, “If you continue this, we’re going to do something to you.” I don’t know what that do something to you could be. I don’t know if it’s you need to have a tech company to go after in Europe that’s strong in the US, and there really aren’t many of those, if any.
I did want to pivot though to a lot of the discussions. EU folks got up there and laid out this big case for growth. You had Macron get up there, you had Ursula von der Leyen do some big talks. I got to tell you, man, I’ll believe it when I see it, okay? And Ursula is the president of the EU Commission. A lot of talk, I’m optimistic. I do believe we need a strong EU is important for national security and interest in the United States. I want to see them do well, but this is the same group of folks. Part of this are Germany, who unplugged their nukes, and saddled up with Putin to pay exorbitant fees on natural gas. And using Greta Thunberg as the poster child for the way that things should work in the future. Not working out too well for Germany right now, so anyways, I’ll believe it when I see it. But good job, Vance, on that speech. Did better than I thought he would. All right, folks. Episode 249, we made it. We were not canceled. This is amazing. So glad to be in town with my bestie, Daniel. I can’t wait for our Valentine’s dinner. I’m not holding my breath to see if you got me a gift, but you never know.
Daniel Newman: Hey, by the way, everybody, it was Pat’s birthday, so send him some tweets and make fun of him. He also became a grandfather this week, and I’ve been hosting this week like the main host. I would’ve spent a lot more time talking about those things, but humble Pat there, because it’s really not about Pat, so we never talk about Pat.
Patrick Moorhead: But it’s kind of about Pat.
Daniel Newman: It’s sometimes about Pat, but it was his birthday.
Patrick Moorhead: Yeah. Anyways, folks.
Daniel Newman: He is a grandfather now. Come on, don’t just gloss over this stuff. It’s more important than anything else that’s happened in this show. And he is my Valentine, and let’s make sure we get that on the record, Happy Valentine’s. And of course, to our wives, to my wife, to your wife, Happy Valentine’s Day. Thanks for putting up with us.
Patrick Moorhead: Take care, everybody. Hit that subscribe button.
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.
