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:
- AWS Summit New York 2024
- Samsung Galaxy Unpacked July 2024
- Apple & Microsoft Leave OpenAI Board
- AMD Acquires Silo
- Sequoia/A16Z/Goldman Rain On The AI Parade
- Oracle & Palantir Foundry & AI Platform
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:
Daniel Newman: Hey everyone, welcome back to another episode of The Six Five Podcast, episode 223. And yes, if you’re looking at your watch and you’re wondering why it’s Thursday and it’s not Friday, it’s because I’m going on workation, but I’m going to try to make a vacation out of it. I am heading out of the country to an undisclosed bunker in a place that no one that doesn’t have social media will be aware of. But if you do have social media, be warned. Bill Ackman doesn’t have anything on me and my intent to be bored and tweeting a lot. Do we call tweeting, Pat? Anyways, how is everybody doing? Pat, how are you doing?
Patrick Moorhead: I’m going to continue to call it tweeting, buddy. I’m doing well. I had some technical issues getting in here.
Daniel Newman: You did?
Patrick Moorhead: Been on the road all week and come in and my PC was not working very well. Hence, the boring background.
Daniel Newman: Do you want to talk about that PC a little bit more? What do you got there, Pat?
Patrick Moorhead: No, this is the Surface with Qualcomm inside. Yeah. I left it in high performance mode, but we’ll figure this out. I did do what I shouldn’t have done. There was a giant update that came in and it was pulsating. You know, that little orange thing in the lower right-hand corner. I pressed it. I did a full firmware update, baby, and this is what we get. It’s like you’re a kid and it’s like don’t touch the stove, don’t touch the stove.
Daniel Newman: But I don’t believe you that it’s hot, dude. I don’t believe you.
Patrick Moorhead: I touched the stove dude, and I got burned.
Daniel Newman: And it’s hot. It is hot out there. But part of our job as analysts is to actually do that. When we break something that’s not normal in how everyone else is going to break something, we’re breaking it because we’re trying. You talk to the president of our second child together, Signal65, the performance lab. Ryan and his team sit around all day and they just try to break stuff. That’s a big part of what we do here. But hey, you know what? I’m just happy to be home because unlike you who flew to New York and did your work and then flew home and got home, I got trapped in New York and it wasn’t fun. So I had the whole awesome experience of doing the Blade. Got an extra TV session in, got across town, 30th on the west side of town, bladed to JFK, got to the lounge, two hours to spare. I would actually even be on time in your opinion, Pat.
Patrick Moorhead: Dan, can you tell everybody what Blade is?
Daniel Newman: Yeah. Blade’s like Uber for a helicopter out of New York. And I want to be really clear, it feels really baller, but it’s really like 75 bucks more than an Uber. It’s not that bad. And it’s five minutes. So instead of a 90 to 120 minute cab ride across town, it’s a five-minute Blade, they put you in a car, drive you right to the door of your terminal and off you go. It’s awesome service. It’s really, really cool. But American Airline, you blow. I hate you. I was in the waiting room and at first thing it’s an 8:00 flight, I’m thinking I’m going to get home at midnight, not bad. Full day of work tomorrow, get that update, 9:30 delay. Five minutes later another update. Canceled. Walked to the front desk. I’m like, “Why is it canceled?” And they’re like, yeah, we don’t really ever know anymore. This is what the American people told me in the Admiral’s stuff. “Yeah, we don’t really ever know. It says weather.” I’m like, “Oh. So that means no voucher, no hotel, no transportation”, but ironically right across town from Newark, there was a United flight at the same time that left and got to Austin.
So apparently the weather was only bad at JFK because apparently Manhattan has multiple climates going on within that 10 mile radius. All right, lots of talking, too much information here. A little bit of a busy week, right after the 4th of July, Pat. We’re back from it all. There was no waiting. On a Sunday I was off, Monday you were off. We were doing Six Five. We did a big Six Five video with the CEO of VAST Data, you and I on Monday. That’s more coming on that soon. We both hit AWS Summit. You were at Samsung Galaxy unpacked. We did another Six Five video there. That’ll be hidden. You did it without me. I mean, they’re always better that way.
Patrick Moorhead: I missed you though. I really did and Brad did ask about you. So anyways. Hey buddy, I think you’re muted.
Daniel Newman: It’s technology, whatever. It’s about to earning season again. There’s a lot of speculation going on, is Tesla going to do its Robotex, we’re not going to talk about that, but that came out. Apple and Microsoft, there’s some news about open AI and their relationship with the board. AMD acquired Silo. Sequoia/A16Z/Goldman are raining on AI’s parade. I can’t wait to talk about that one. Then Oracle and Palantir made an interesting announcement. And then Pat, we’re going to talk about legalizing voting with or without IDs. No, I’m just kidding. You got to go to our Twitter accounts to get that kind of stuff. All right everybody, do you remember though this is for information entertainment purposes only and while we will be talking about publicly traded companies, let’s not take this as any investment advice. I’m going to kick off the first topic, Pat. This is going to be probably the beefiest one we’ll have, although there is a lot to cover and that’s going to be AWS Summit.
So look, a lot of content happened, and there was a little bit of a prelaunch because the Graviton for CPU was announced. It came out the day before I came down with a little illness. I was in my room watching Matt Garmin on CNBC talking about it. Great to start to see him get out there by the way, look forward to having him on the show. And look, AWS did the most AWS thing that AWS does. They launched a new chip. It’s got some great improvements on generation to generation and they really only focused on that. They didn’t talk a whole lot about how Graviton compares to other merchant silicon that they sell. They talked a lot about, and really efficiency was the big focal point of this announcement. They talked about, I believe it was 60%, Pat, more efficient than its previous generation and they also talked a lot about the ability for this next generation to reduce carbon footprint. Now they did make a reference to non Graviton instances. That was about as competitive as I’ve ever seen, but like I said, the topic was all in on efficiency. They did the exact same thing by the way, as they ran through second generation Inferentia and Tranium chips, talking about two times more efficient out of their new Tranium chip and 50% more energy efficient on deep learning. Now again-
Patrick Moorhead: Hey Dan. Hey Dan. They announced this chip a long time ago. This announcement was about GA.
Daniel Newman: Oh, yeah.
Patrick Moorhead: No, no, I’m just doing the check.
Daniel Newman: You’re talking on Inferentia or Tranium or Graviton?
Patrick Moorhead: Graviton 4.
Daniel Newman: They announced. Yeah, I’m sorry. I should have said that clearly.
Patrick Moorhead: No, no. For the audience.
Daniel Newman: It came out . The tour, they were on announcing it’s out, so I should have used that word more clearly. They announced that it’s out and it’s available now. My point though is really that they were doing a victory lap on efficiency. They seemed to double down, lean in hard into the power envelope of these AWS homemade. I appreciate you stopping that. It probably came across like I was saying, they were announcing these chips for the first time. GA, everybody, GA, Pat. Thank you. All right, let me talk a little bit more about the bigger AI summit or AWS Summit focuses and that was the Matt Wood keynote. Matt Wood, a VP of AI services, he came out and I did a long thread, lots of different things that he was talking about. I’m going to kind of try to focus on a couple of things-
Patrick Moorhead: The way Matt Wood did a Six Five summit keynote on end-to-end enterprise AI.
Daniel Newman: He did, and you almost worried me there that I’d given the wrong name that I’d also botched that. So thank you.
Patrick Moorhead: I’m not punching, I’m, just clarifying for the audience who doesn’t follow it. I know you knew it was GA. It was-
Daniel Newman: Yeah. Fact checking department. Well, but you know how I am though, Pat. I acquire a company and I announce it and then when I actually close it, I announce it again. I mean is this any different?
Patrick Moorhead: No, that’s exactly what AWS does. It’s smart.
Daniel Newman: Yeah, it’s smart. And they were smart. They did a whole nother media round and got a whole nother tour to talk about their launches. So anyways, point though was a couple of interesting things with Matt that he came out. One is he really started off big on wanting to establish the footprint that AWS is in the AI business. Let’s talk about where that happened. They ran the stat 96% of AI ML unicorns run on AWS. Where have we seen that stat before, Pat? Maybe it wasn’t 96, but it was some version of ninety-something that has come from every-
Patrick Moorhead: It was 97% or something,
Daniel Newman: Every other provider. So it’s a really interesting thing. But what I’ve come to the definitive conclusion is that multi-cloud is a big part of every AI strategy. They also mentioned 90% of the AI. One thing I did really like though is that they talked about the development and build out of their services, Pat. They talked about having 326 new services that were launched, two times as many as they’re saying as their nearest competition. Now AWS got a little goofy with this. Not goofy. They got more, you know how they’ve always been so PC with this. They actually did a slide Pat where it showed the AWS list of services and then right below it’s a company one and it was a color, red, green, blue, yellow, and they were talking about it was more than two times as many in the last 18 months. I mean, is this a little different feel from Matt, under Matt’s leadership?
Patrick Moorhead: They’re getting aggressive for sure and there’s a perception that they’re behind and –
Daniel Newman: Yeah, and I like that you said that. I’m going to go quick so you can get to your stuff because there’s a lot here. I’ll be really a bit succinct with the rest of the stuff. I like that they pointed out training foundational models on AWS. There’s only so many places this is happening, but some of the big ones, Anthropic, Mistral, Hugging Face, AWS, AWS, AWS. They also did some very exclusive fine-tuning with CloudThree. CloudThree is getting a lot of very important attention for its accuracy, its fidelity, its capabilities. So there’s a deep partnership with Cloud and obviously Amazon being a big investor in Anthropic. Last thing they did a how many LLMs data point. I really like this. For everyone out there, enterprises are not going to standardize on a single LLM. Let’s be very clear. Most are using two, three, four or more. They’re going to use multiple clouds. They’re going to use multiple LLMs and Pat, don’t be surprised if they use multiple silicon architectures to get this done.
Patrick Moorhead: That’s it. You done?
Daniel Newman: I had a lot more, but we’re like 30 minutes in because you were fixing your computer for the first 22 minutes of the show.
Patrick Moorhead: I know. I’m sorry everybody. There was a lot more goodness I know Dan had on that. And yeah, by the way, I participated in a, I don’t know, was a 10-minute video that Yahoo Finance did on Graviton four and gosh, my tweets were in there. I sent in photos, I did some video for them as well. So it was fun I got in on the action as well. So let me boil this down into two things. What AWS brought out here was either better results for enterprises or increased efficiencies by developers. On the better results, you had more data sources, RAG connectors to Salesforce, Confluence and SharePoint, not just the data that’s sitting in S3 buckets. They added guardrails for hallucinations and that’s about again, getting better results. For agents, which is a new way to do an application, which is essentially an agent talking to an agent, talking to an agent and sometimes even creating a code on the fly.
And what they did is they added memory to the agent. In a very similar way, if you use open AI or perplexity, it has a certain memory and it just gives you better results. And on the increased efficiency, they added Q developer to SageMaker. So SageMaker is the end-to-end data and AI platform all the way to data ingest to training to deploying the models. And then of course Bedrock sits on top of it and connects it to create enterprise applications. So you can actually go in and do what you would expect, which is create code inside of SageMaker. You can have it write code to create a script to do data cleanup as an example. So exactly what you would expect Q. The competition has co-pilot, Google has agents, what you’d expect it to do. Also, basic how-to manuals, how do I do this?
And it goes in and it tells you exactly how to do something. And the second thing they added is adding to developer to your own code base with the ability to customize that as opposed to the code that you wrote that is essentially housed and new on AWS. And the third one was what’s called App Studio that went into preview and that’s building applications with natural language. And we’ve seen examples of this out there out from many different vendors to be able to do that and AWS is bringing it to their capability and one thing I do want to point out is that AWS, while the company has a consumer play with .com and the media and all the different devices, this stuff has to be bulletproof and cannot be put out there. The expectation is that whatever they put out there is absolutely bulletproof and overall I did sense an increased aggression.
I do believe that they are feeling some of the heat. There’s a couple of things going on here. Microsoft and Azure, two things. First of all, there was a lot of people talking about OpenAI and the capabilities that were put out there and writing on top of Microsoft. Microsoft came out big time on the consumer side with Bing and Bing with Copilot and then Consumer Copilot that is on every modern PC out there that’s updated. Microsoft also did the cheat code and I don’t mean that in a bad way is they partner with OpenAI and then you’ve got Google with doing all their training on the TPU. Bard one and Bard two, didn’t come out very well.
Even when Gemini came out, had some mistakes and some issues, but when it was running and it is much improved almost every Google search you can see some of the generative AI capabilities. AWS doesn’t have that capability yet to get all those people talking on the consumer side. So yeah, it’s going to be very interesting. AWS, its posture is being first. They’re the first one who stood up cloud computing and this is a very different area for them and it’s interesting to say the least.
Daniel Newman: Yeah, there’s a lot there, Pat, and I mean, we could talk more and more about Bedrock and how they’re building a very open sort of platform with lots of experimentation capabilities. I mentioned services. What the real question I will ask as we head into this next era of cloud is how much does that established market leadership and infrastructure translate to the AI era? How much do new disruptors like CoreWeave, Lambda and others impact it and how much does the app layer being more present in some of the others with Google and Microsoft really impact the AWS business if it doesn’t all exist because AWS has never quite, but Q is really interesting. Anyways, we got to move on. Lots of good insights there buddy. Let’s talk about Samsung. Unpacked. You were able to bounce between two things in New York. Did you get to Paris too? Did you do that time travel?
Patrick Moorhead: I was originally going to go to Paris but decided not to, but I ended up doing three things in New York. That was pretty awesome. But gosh, there’s so much content here and I’m going to have to blow through. But Samsung brought out a health ring, a completely revamped exercise watch. They brought out their sixth generation of the fold and their sixth generation of the flip and they brought out a third generation of their what’s called buds or their in-ear speakers and they brought out a ton of new AI stuff. So some of the highlights, so Z Fold six, I can’t believe how thin this thing is and how good the multi-windowing and the camera is getting. There’s a lot of chatter about, oh wait a second, you increase this by a hundred bucks and you made all these improvements but oh, it’s not enough, I wanted more camera.
I see the fold as a B2B play and quite frankly, it’s the killer app and it’s the killer device for hardcore business gets the new Qualcomm Snapdragon, Snapdragon eight gen three, which you get all the AI goodies, CPU GPU, huge bumps there. Flip Six, I mean it’s amazing. You put the flip six when it’s open and you put an iPhone or even the Galaxy S24, it gets tough. I mean they’ve improved the camera. They have really narrowed the gap on performance, battery size, battery life, wireless and durability, and I love these fricking cute cases, heavy target market to a younger audience. I think they really have something there and there’s going to be any flip phone that will popularize this and do high volume. This is where it’s going to start. The watch, I’m really into health and there were definitely some differentiators versus Apple Watch, first of all. Sleep apnea detection, something that’s called advanced glycation and products. Again, I’m doing a lot of research on this, but essentially gives insight into metabolic health, metabolic aging, and typically these things are done with a blood test. Also, ECG monitoring and blood pressure, unfortunately not for the United States.
They’ve had this capability for a couple of years, but the FDA is just not moving forward. Interestingly enough, in many countries in the EU it has been certified. But what I’m really another here is you get what looks like, and again, I need to use this a very valuable app you don’t have to pay for it. I’m paying for a ton of applications because Apple doesn’t actually embed health capability that tells you what to do. It gives you the reading. The only thing that the Apple Watch does natively, it tells you when to stand up. Okay. I mean literally that’s. It doesn’t say where you’re overtraining. Undertraining here is an issue with your sleep that you need to look at. It’s very basic. And one might say, hey, they’re making way for their developers, but with Samsung, you get it all there. Now, can’t vouch for the app yet. I haven’t used it. And then there’s the ring. I’m a huge Aura Ring user and I primarily use it for deep double click on sleep and a double click on stress. Some differentiators here. It has ECG functionality, it has continuous heart rate monitoring, and it has this thing called a vitality score, which is very similar to a paid application that I use called Athlytic. That tells me essentially my physical and mental readiness.
Daniel Newman: Do you ever allow that to particularly influence your day when your score isn’t as good?
Patrick Moorhead: No, no. We’ve talked about this, Dan, what it is more of a, hmm, gosh, I kind of feel lousy right now and let me pop open and look at it. It goes both ways. It also, it might be all green and it lifts me up, but over the long term, my watch plus my ring plus an app called Athlytic have been instrumental in my health turnaround.
Daniel Newman: Is that it?
Patrick Moorhead: No, no, just on the AI side. Again, Samsung announced… Oh by the way, Samsung has been shipping their AI stuff for months, like circle to search, sketch image and stuff like that and-
Daniel Newman: They’ve reiterated GA.
Patrick Moorhead: Yeah, this is… No, it’s actually been GA for a long time, these features. But what I’m going to talk about are some of the standouts here and some of the new features. You’re not going to challenge me on this one, man, you didn’t go to the event.
Daniel Newman: I’m not even challenging you. I’m just asking dumb questions to keep the-
Patrick Moorhead: They’re good questions. No, but they did come out that circle the search is the most popular AI feature that they have. I think the sketch to image, the browsing assist where you can essentially, you hit your browser and you can summarize page, one of the most interesting was PDF overlay translation where you don’t even have to take it into a PDF viewer and then do the trans… It’ll do it right on top of that. I thought that was awesome. And if you look at the flip and the fold, the ability to do translation where you can be speaking into your phone and then the other person can be reading it in their language is pretty amazing. And then side by side, if you have a meeting going on and it’s translating on the right of the fold screen, I just think that’s incredibly powerful. So I could talk about this forever, especially the health features, but I have the fold and it’s super thin and I’m not even going to put a case on it for the first time ’cause it has squared edges and they made it wider. So I think it’s good stuff. I like it.
Daniel Newman: Yeah, I wasn’t there. You hit it pretty hard. I haven’t fully read up on it, Pat. I think there’s an interesting sort of AI challenge that’s going to take place between the two platforms. This isn’t just really Samsung ai. It’s going to be what does Android powered AI look like? And of course Samsung’s always the de facto beneficiary of a volume. So Samsung AI plus-
Patrick Moorhead: Yeah, Samsung plus Google takes on Apple plus OpenAI.
Daniel Newman: There we go. But I mean of course OpenAI is not really embedded. I mean it’s nothing more than A, I still don’t really get that one, Pat. I got to be honest with you and don’t challenge me. I actually do get it, but I don’t get it because it’s like the ethos of Apple and we’re super private, but yet we’re going to embed this company in the OS that nobody trusts anymore. Speaking of no one trusts this company anymore. Let’s talk a little bit about Apple and Microsoft deciding to, well, in Microsoft’s case, deciding to leave its role as an observer on the board of directors at OpenAI and Apple deciding not to pursue what was expected to be their participation, following that deal, to be part of the observers of the board of directors at OpenAI. Man, Pat, look, that company builds the coolest technology on the planet. OpenAI is doing things unimaginable, but I still tell you there’s something going on. It’s like a volcano in that thing and it’s just burning from the inside out.
He casts a controversial figure. His top lieutenants are all leaving. One of them is starting a company that’s literally got the name of your ethical or whatever he is calling it AI in the name of the company. And to me that felt like a very direct dig at the fact that OpenAI’s existence was not based upon that. I did see an interesting, Jay Cal from All In shared a video I saw of Sam driving around in a two and a half million dollar Koenigsegg to which he said, “I will be starting an NFP or whatever. I want to start a non-for-profit or whatever.”
Patrick Moorhead: Yes, exactly.
Daniel Newman: And look, I’m a capitalist. I’m all for people being successful and what Sam’s doing is changing the world, but Apple and Microsoft increasingly want nothing to do with it, which is weird because Apple has obviously wants a lot to do with it, but they don’t want to be uniquely close in a way that the outside world can see. So I’m going to give you the caveat there in a minute. Microsoft, even moreover, and there’s been that controversial commentary that’s come in the market from Satya Nadella talking about them being over and under and all around. And basically no matter what Open AI does, they have the whole thing under their control. But being observer on the board seems to be that cutoff limit of where they do not want to be anymore because guess what? The probers are busy probing again and they want to figure out if these companies are colluding or doing anything that’s going to uniquely position and give advantage.
And would the spot as board observer give them more power? Here’s the thing, with them being the largest commercial relationships, not money necessarily because the Apple Open AI one doesn’t have a direct monetary relationship, at least not disclosed, but with the influence of money I still think that both those companies are going to have an outsized ability to influence where and how open AI goes directionally. And of course we’re even hearing Open AI potentially changing its corporate structure to enable it to go public or in some way be able to be a beneficiary corp that can in fact be invested in more easily by more outsiders. The only thing I want to add about this is if you remember Microsoft made that inflection deal with Suleyman, and I think a lot of the company is AI strategy revolves around that. They are more directionally trying to build stuff on their own. They basically acquired inflection, but they don’t call it an acquisition. They did it that way very intentionally. They were able to avoid all scrutiny, all regulatory approvals, but they did basically hired everybody, licensed all the technology and left nothing in the company other than a shell.
So it was really interesting, by the way, super smart, super smart. I want to be very clear about that. But now the regulatory bodies are probing this. They’re like, “Is that legit?” And I think we’re all saying, “Hey guys, waking up. Good morning. How’s it going over there in government? It’s good that you’re starting to pay attention.” So the long-short Pat is they’re all going to be involved. They’re going to continue to use, they’re going to continue to influence. They just don’t want to be named as formally participating in the shaping and driving of the company’s governance.
Patrick Moorhead: Yeah, I don’t have much to add on this. It was interesting, your first hack that you were going down I was wondering if you’re going to hit the regulatory piece, but yeah, I mean both Apple and Microsoft got word from either FTC or the DOJ and they had to bail and it was amazing. I think it was on the same day was just so shocking because I think everybody was surprised. So Microsoft got put on the board after the big shakeup in leadership there. Sam was CEO, he got fired, then he was going to be a Microsoft employee, and then he came back on and then Apple just received their board seat after the deal that they did for Apple AI. So yeah, I think this is pre-crime and these guys are getting ahead of the pre-crime to not get these folks on their back. I haven’t read the filing or the warning, but it must be a lot of fun.
Daniel Newman: So talking fun, acquisitions are fun. AMD doesn’t make as many as some companies, but when they do, they tend to make decisions that are substantial. And they made the acquisition of Silo this week, Pat. What’s your take?
Patrick Moorhead: Yeah, so about a $650 million acquisition, big enough that they had to float this out there, 300 employees, 130 of them are AI PhDs and here’s the TLDR. AMD is going to use Silo for two things. The first one is to optimize LLMs for AMD hardware and even machine learning, object and voice, AMD had to do optimizations of those frameworks. And for LLMs, they need to do the optimizations for those. And the second one is to really help end customers either optimize, tune or prune models for their specific use cases. And Dan, as we’ve seen just in all our travels and all the briefings and all our discussions, you can take a model and then you can try to get the results that you want through the data that you put in front of it. You can actually go in and customize the model and fine tune that model to see what you get.
And then in some cases you have to create your own model if it’s very specific or it’s not one you can pull off the shelf and license from a Hugging Face.
So this is very much what NVIDIA gets a lot of credit for. When model creators for training create a new model, they do it first typically on NVIDIA, with the exception being Google, Gamma and Gemini and AMD has to optimize these models for their hardware. And then even out in the field when there’s an end customer that needs help in again optimizing, tuning and pruning, this is where the silo folks come into play. Not bad. 300 employees, $650 million, Dan.
Daniel Newman: Yeah, I mean look, it’s a big spend, it’s a big turn, but I mean it’s also a big bet. And the risk reward profile is pretty substantial, Pat. This talent, it’s not just like you can run over to the local university and find this 125 PhDs and these people that have all this experience developing, building, tuning and running the models and with AMD experience too. So to your point, a lot of the talent that’s out there is primarily skilled for NVIDIA if it exists at all. And they’re very busy. The people that are out there that know how to do this right now. This is a case of AMD trying to hack the speed. When I say hack, I mean it in a positive way, the speed to which they can develop software to run on instinct. This is the biggest challenge the market has when it comes to AMD.
It’s never is there hardware good enough? In fact, the benchmarks stand up pretty significantly and amply at this point in time and it’s actually put some pressure on NVIDIA to launch its newest models. But having said that, the software is always the part that people come back to and say it’s just not there. It’s not as complete, it’s not as ready, it’s not as utilized. There’s not as much development talent. And so AMD in this case is acquiring a lot of talent that can help them foundationally move quicker. And I mean that as a double entendre, both foundationally in their business and foundational models. And then they can actually connect the stacks and develop applications that people will use. Because that’s the other thing, Pat, we’re going to talk more about that in a minute. But the big problem of AI right now isn’t hardware.
It’s not an infrastructure problem. As much as there is some accessibility to GPUs and some network build out, we are in a massive infrastructure build out. But now people want to know how do we use this stuff? Well, AMD needs to, if they want to win, they want to win more business, they want to take a bigger share of this market, Pat, $200 billion in CapEx has been spent so far. AMD has gotten what, four or 5 billion of it this year. So where did the rest of that all gone? They want a bigger piece of that, they got to make it more compelling. This is part of that. So $650 million to get a bigger piece of what I think Lisa called out as a $400 billion opportunity by the end of the decade seems like a good bet. And she makes mostly good bets. So speaking of betting and problems, this past week, this has been my new hill to die on, has been an endless call from market makers and market influencers like Andreessen Horowitz and Goldman Sachs and Sequoia.
Sequoia published a piece called AI’s $600 billion problem or challenge. A16Z did a version of this over a series of a few blogs. Goldman Sachs basically came out and said there isn’t a single application on generative AI that basically delivers any value, more or less. The words, those are paraphrases, but those are the words. And so I’ve been thinking a lot about it. I wrote an op-ed, I wrote a few tweets that could have been op-eds. We’re doing a lot of market assessment right now, market growth. If there is $200 billion in CapEx and we need $600 billion in market impact, how does this work? And so I’ve been talking about of these network effects and basically this $200 billion has all been seen by the compute memory systems network. And then of course there’s some power and other elements that have gone into this build-out.
And then you’ve seen where’s all that spend going and that’s where a lot of red flags, people have been somewhat critical of Google, Microsoft, AWS, this wave of startups I mentioned CoreWeave, Lambda Labs and others that are spending a fortune and trying to understand what services are they delivering that people are paying for to justify those huge expense outlays. And then Pat, really the big part is we’re hearing terms now like is software dead or software as we know it is dying? And that’s basically all the ISVs saying that Salesforce, UI path, Workday, ServiceNow, SAP, parts of the Oracle stack, Snowflake, go down that list and say is it the end for them. I mean they’ve been crushed in the market. So there’s been all this excitement and exuberance and enthusiasm about tech’s boom. It hasn’t been all of tech. It’s been like seven companies.
And so a number of companies are being left in the dust with software being a big part of it. But here’s my thesis, Pat. The $600 billion comes from a third network layer that is effectively analyst firms, manufacturing companies, automotive makers, financial services providers, healthcare firms, investment and financial services banks, and they’re going to be using apps, Pat. They run custom software and they run Oracle, they run SAP, they run Salesforce, they run a whole boatload of workloads and apps and guess what? They do a lot of it in the cloud consumption. They do a lot of it in applications, the ones I mentioned. So what we have is an ingestion and then digestion problem. We’re ingesting all this, we’re digesting all of this and basically we’re looking at a time horizon. And so my ultimate thesis here, Pat, is that there isn’t a $600 billion problem.
There is a timeline problem, but if you’re in the market, the timeline problem is more substantial than if you’re in the business because in the market you’re trying to time a trade, you’re getting in, you’re getting out. Am I buying high? Am I buying low? When’s the growth going to come? When am I going to be able to see it in the numbers? One of my favorite lines from one of my favorite movies was called The Big Short, and he walks into the office and he says, “I may be early, but I’m not wrong.” And the guy in front of him says, “It’s the same thing.” It’s the same thing. So if you’re a trader, you are possibly early. Software may be slow to be able to define revenue and then discern between what’s growth of revenue and what’s commodity, Pat. Do AI PCs and phones mean more revenue over the long run or do people just expect more features? And that’s the expectation.
Patrick Moorhead: Oh, you’re on mute again buddy.
Daniel Newman: The other side of it is with SaaS it’s the same thing because the whole SaaS concept was it was born on continuous improvement. So when do people pay for generational quality leaps in software and when is something just supposed to be a baked in feature that’s better. All right, I’m let the topic lie here. Pat, I think if you are a builder, you’re very excited and enthused about the future. I think if you’re trying to trade this, it’s hard to get away from those first network effects because it’s not obvious yet where money’s going to be made.
Patrick Moorhead: I believe this is a hundred percent about timing and it’s hard to see the beauty of the wood finish when you’re putting the plumbing in and building the cement foundation. And when it comes to the infrastructure, that’s exactly what we’re doing and we’re actually having to train the home builders on a new way to build a home and new way to install the cabinets on there ’cause it’s never been done like this before. The one element and maybe you hit, maybe you didn’t, I didn’t interpret it that is competitively. I mean, what is a SaaS provider going to do? Just say, “You know what? I am deciding that I’m just going to not do this right now because I don’t see this thing panning out for another two years”, and all your competitors go in and they institute all of these features and then you’re going to lose all of this business and your stock’s going to get dumped and you’re going to get fired as a CEO and a leadership team.
So there actually is no choice. I think it is a miracle at how these SaaS companies are walking on this razor’s edge of not bringing in a ton of cost and slowly and quickly bringing these features out. One thing I know though, like Adobe Firefly and the ability to cut your content creation costs by 50%, the ability in human resources or any type of service environment to improve self-service in a way we’ve never seen before. I mean, it is just crazy the amount of code that can be written. I mean, heck, agents create code on its own now it’s Sandboxed first to keep it from doing weird and nefarious stuff, but it’s there.
So it’s a timing issue. If you’re an enterprise writing your own enterprise applications and capabilities or you’re an enterprise SaaS companies, what are you going to do? You going to not do e-commerce? Are you not going to put up a website? You not going to connect to the internet. We’ve had these types of milestones before and I know there’s been, listen, I was 10 years into my career when .com boom hit and the bust hit and I think we’re so far away from that. I mean, the internet people, there wasn’t good content on there. I mean it was like you get on and you do search and the search sucked. Why? Because a lot of the content wasn’t even on the internet. I am actually quite optimistic and quite bullish. I will let the financial bean counters figure out the timing and the affordability, but you don’t have a choice if you’re an enterprise or an enterprise SaaS company, you got to move.
Daniel Newman: Yeah, you’re getting into it. And that was what I was saying about what’s expected Pat or required to stay competitive versus what can you actually incrementally charge more for is what I was trying to allude to is of course –
Patrick Moorhead: Everybody but Zoom and a couple other companies are paying. It’s an extra added feature.
Daniel Newman: And so when will people pay for it and do they get enough value to justify? These companies have always just raised prices anyway just for doing iterative feature growth and expansion. That’s the whole model of SaaS, Pat. Now, one thing I know you love to talk about, and we’re going to get to the last topic here, but you’ve always loved to talk about that hybrid multi-cloud data fabric. And one of the big things that the Goldman person that wrote the piece and his name I forget right now, but is he pointed out that he could have it do some of the complex equities modeling for him, but it was at six times the cost. And so what I think a lot of that has to do with is the initial development of a model and training and bringing all the data together on a fabric that can be appreciated by a generative tool is going to be front-loaded expensive, so at six times or eight times, whatever it was.
But over time, once you’ve optimized that, there has to be a factory element to it as you spend big to build it. And then over time it starts to create lots of efficiency. And I don’t know that they’re weighing that in yet because right now it feels like it’s all build and there’s a continuous build because the data. But listen, 20 years of ML pipeline, Pat and everything else, we still haven’t gotten that multi-hybrid cloud data fabric to where it’s easy to use for a real looking enterprise data state and that’s going to slow down ai. We got to figure that out. It’s got to get figured out. I know there’s some things out there. Probably it’s something we should have a whole different show to talk about, but let’s talk about actually a company that’s trying to sell AI and ML and Data Pipeline, Pat, and a new partnership that’s going on between Oracle and volunteer. Expanded partnership, I guess I should say.
Patrick Moorhead: Yeah, so two very interesting companies. You have Oracle that is growing like crazy as an IaaS provider, complimenting their SaaS and PaaS capabilities. Oracle has deals with Microsoft, they have deals with OpenAI. I mean it’s crazy how competitive that they are on IaaS with this Gen 2 OCI and then doubling down on all of the GPUs. And then you have Palantir. It’s interesting, Palantir formed in 2003. Peter Thiel founded and created and it really specializes in big data analytics for big defense, big government and also commercial interests out there. And they have three main product lines. They have Gotham, which is used by defense and intelligence agencies. They have Foundry, which is all about for commercial enterprises. And then Apollo, which is essentially one step underneath that that is managing the CI/CD pipeline for these companies. And it’s funny, you really had to look at the press release to see what was going on here.
Hey, we’re certifying Palantir’s platforms on OCI. Palantir is moving Foundry workloads to Oracle to OCI. And my question, everybody is from where? Nobody wants to say where Palantir was doing this before, but this looks like an absolute net add for OCI and a net lose for another IaaS provider or that they were doing it at a colo or something like that. The other thing we found, and this makes sense with Gotham, think defense and intelligence agencies, that application and AI platform is distributed across all Oracle cloud regions, dedicated regions, Alloy, Oracle, government, Oracle Roving Edge, pretty much everything, air-gapped for defense and intelligence implementation. And this just seems to me so classic Oracle and right in their sweet spot. So to me the TLDR is another big IaaS provider, another big IaaS win, and then a distribution of applications, in this case Foundry and Gotham available to Oracle customers.
Daniel Newman: Yeah, I’m going to hit this one quickly, Pat, because I think you made some good points. I don’t think there’s a ton of meat on this bone, but what is the meat on the bone here is there is such a symbiotic relationship between one of the rising stars of being an intelligence and data platform to government, to top secret, to your highest security agencies and clearances being Palantir and of course Oracle, which is basically the database that’s run every government entity on the planet. So the go-to-market motion here makes a ton of sense. It’s so logical that this is a build on, build up. And of course Oracle, while they are building AI capabilities and databases, and obviously they are not building out as expansive of an infrastructure or a software platform as some of the competitors.
So Oracle’s approach using partnering and best of breed to be able to corner a market like the intelligence market, the government defense market, so successfully makes sense. I mean, Pat, there’s an argument that at least within most sovereign and government that this is going to become a de facto standard. And I don’t know who else is going to be able to crack this from both a clearance, a certification and an already an existing platform within these different agency models, so –
Patrick Moorhead: Yeah, I looked across AWS, Google and Azure and all three of those companies have certain alignments. Google is more about commercial applications that you can run out of their marketplace. This was the most like, hey, defense, air gapped, intelligence. Just in your face and proud.
Daniel Newman: And I mean, look, that’s a big market. It’s a stable and sturdy market. Remember there’s a company that has 74% more predictable revenue at any given time with a big cloud growth, big infrastructure growth.
Patrick Moorhead: Yeah. I said this gosh, two, three years ago. They have the most straightforward private cloud and sovereign cloud offering out of anybody. It’s just straightforward. Their on-prem cloud as essentially, you take a copy of what’s in the public cloud and you paste it on the customer’s premises or colo and you put a chain link fence around it and Oracle manages it for you. And then Sovereign Cloud is a different building in a different… There’s no, oh, oh, wait a second, wait a second. And I think from a market timing, Oracle got a jump on Sovereign over everybody.
Daniel Newman: They did. It’s sticky. And this is a good partnership. And by the way, Pat, I don’t know if you know this, but Palantir quietly saw its market cap exceed Snowflake.
Patrick Moorhead: Interesting. And it’s crazy that Snowflake what formed in, I don’t know, 2020 or something, and Palantir’s 2003.
Daniel Newman: Yeah, they’ve been at it a long time, but nobody has really seen them as a big, fast riser. They’ve been big, fast riser and their growth is much faster than other software companies right now. And their logos have been really significant and the growth of companies that are running its platform. So it’s a good win, good partnership. Solid. Good stuff, Pat. All right, we got to wrap this show up. We did it. This is the prelude to a vacation. We won’t be back next week. Pat can do the show, but it won’t be with me. Maybe it’ll be with me. I don’t know. Maybe I’ll get bored. Maybe by the end of the week I’ll be like, Pat, I got to get on. I got to do something. Who knows.
Patrick Moorhead: Do it from the beach.
Daniel Newman: But don’t count on it, everybody. I know you sit around, you plan your weeks around the 9:00 A.M. Friday, Pat and I. You can’t help yourself. Listen, everybody hit that subscribe button. Join us for all of our coverage here on The Six Five. Pat and I love doing this show. We love the community. We appreciate each and every one of you that tunes in, but we’re going to say goodbye for now. We’ll see you all really soon.
