Skip to content

Podcast

2024 Year-End Review and 2025 Predictions

The Cloudcast

Source ↗ ← All highlights
  • Rapid Evolution
    • The hosts analyze the major cloud and AI news of 2024, noting the rapid pace of change.
    • They emphasize how quickly things evolve, particularly in the AI field. Transcript: Aaron Delp He’s not just saying. He could afford us. And he might be the sort of the poster boy for, you know, things can change in a year. Be careful. Be careful what you wish for. So what I did was, and I don’t know if we’ll put all this in the show notes, we’ll put it in something that’ll be linked in the show notes so folks can see this. I kind of went through and pulled out both on the cloud side and the AI side, some big things that happened in Q1 through Q4. So we did it on a quarterly basis. You know, we’re not going to go through all these, but we did want to kind of do is like Aaron said, stuff feels like again right now and partially it’s I think because one one ecosystem Is sort of stabilized in terms of cloud so those things that change aren’t as big and then you’ve got this new sort of ecosystem trend with AI coming along which you know crazy stuff’s Going on lots of money’s pouring in and so we’re going to kind of contrast those from quarter to quarter just to sort of show what happens in a year, especially if you’re new to this world And you’re like, huh, is this normal that things kind of change fast? It ebbs and flows. It depends from year to year. So this will be an interesting exercise. Why don’t we start with Q1? And I’m going to do this. We’ll each take a quarter and you can pick either cloud or AI, kind of pick, you know, kind of paint a narrative for the quarter and then we’ll do a little bit of commentary around it. (Time 0:03:54)
  • Q1 2024 Challenges
    • AWS struggled in Q1 2024 with an unclear AI strategy, eventually investing heavily in Anthropic.
    • Broadcom’s VMware pricing changes caused disruption, and the CNCF faced an existential crisis. Transcript: Aaron Delp Yeah, absolutely. What do you want to start with? Brian Gracely Let’s do Q1 cloud. OK. Go ahead. Go ahead. Aaron Delp No, I got it. Brian Gracely I got it. So like, okay, the biggest thing I would say right now, and this, you know, led to big changes, the AWS and the strategy and the AI strategy early in the year, right? You had Celepsi and then everything that happened with him. And then the AI strategy was very lacking. And then they, it felt a little bit like the knee jerk reaction of like, hey, let’s go invest a whole bunch of money in Anthropic. And then, Oh, by the way, and we’ll announce it as we move to Q4, like, Hey, we got our own models now. And so it was, it was very noticeable that AWS’s AI strategy and just AWS in general was, was struggling in Q1. And then Broadcom really they announced the big changes to the VMware pricing and that led to a pretty disruptive year to, I’m willing to bet everyone that listens to this podcast. The CNCF, there was a bit of a mix up of like, hey, there’s some stalwart companies that are kind of keeping a bunch of projects going. And then what happens when some of those projects change? And can a project go from incubating to mature to back to incubating? Like there was a whole thing going on with all of that. And that’s where just this, you know, open source companies and open source strategies and rug pulls was introduced, you know, it started in the crypto market, but now we had open source Rug pulls. (Time 0:05:11)
  • Q1 Uncertainty
    • Q1 2024 revealed uncertainty in cloud and AI.
    • AWS’s strategy was unclear, VMware’s licensing felt like “Oracle 2.0,” and the CNCF’s stability was questioned. Transcript: Aaron Delp Yeah. Yeah. So I think that’s a good that’s a good starting point. We started with cloud and cloud felt like, you know, the couple of pillars that people knew a lot about AWS felt a little bit uncertain for the first time in, you know, a decade plus. Yeah, really ever. VMware felt like people were like, okay, this is what the new thing looks like. And it felt weird, right? Because it went from being a sort of your friend, you know, sort of your friendly company that did interesting technology to, you know, oh, this feels like Oracle 2.0, you know, licensing Changes. And then, yeah, the CNCF was starting to face a little bit of existential crisis of, well, you know, are things stable and are, you know, are projects stable, are companies that are supporting The main things stable and so forth. So interesting start to the year on the cloud side. On the AI side, it was a little bit of uncertainty. So we were coming out of the original thanks quitting, thanks firing with Sam Altman from the year before. Google immediately kind of releases, comes out. So chat GPT4 is out there. Chat GPT is kind of making waves and they come out with Gemini 1.5. So it was sort of their first time of being like, here is this commercial, you know, big model, you know, big, huge, super capable, um, you know, GPT four, like, uh, LLM that was out there. And, and Google went through a few little weird things at the beginning of the year where it was like, some of the stuff was felt like it was a little bit, I don’t know, I don’t want to say Censored, but like there was some weird stuff they did in there. But they’ve been generally kind of straightened themselves out as the year went along. We started to see Microsoft have sort of the second order thing started to happen in terms of like, what is your partnership with OpenAI, right? They started having partnerships with other companies. NVIDIA became the most valuable company in the world by stock value. So all of a sudden, you know, the little gaming company that made, you know, GPUs and so forth was now the biggest company in the world. You know, Jensen had made his, has done his world, had done his world tour the quarter before for everybody’s trade show. And then, you know, we started seeing the beginning of a lot of companies were selling their data to the LLMs, right? So we ended the year with people being like, there’s not enough data left. We started the year with every data source, you know, whether it was Reddit or whether it was newspapers or whether it was other things being like, sure, we will monetize our data to these Things. And that was what sort of seemed like a big deal at the time we got towards the end of the year. And it was like, we’re we don’t have enough data. We don’t you know, like it was that sort of stuff. So it was interesting to sort of watch. You know, we saw a lot of responses early on. Oh, the last thing that was happening at the beginning of Q1. And this I think we’ll continue to see, although it’ll be interesting to see how it’s tempered. This is where we’re starting to see people talk about trillions of dollars of funding being needed, you know, completely rethinking. You know, this was this was the beginning of sort of Sam Altman kind of painting this picture of 20 years in the future. You know, what type of investments were going to be needed. AI was going to be at the center of every part of everybody’s life, you know, these sort of like crazy grandiose things. And people were like, that’s a lot of money, right? That’s, that’s gonna, so yeah, that’s sort of where we started the year. An interesting, interesting place to start felt like a little bit of, of, you know, shaky grounds in Q1 in a lot of places. (Time 0:06:53)
  • Q2 Acquisitions and Earnings
    • In Q2, IBM announced its intent to acquire HashiCorp, and cloud providers’ earnings returned to normal.
    • AWS stabilized with a new CEO, and AI saw acqui-hires impacting employee payouts. Transcript: Aaron Delp And that got folks a little bit concerned about things. Yeah. Brian Gracely I mean, there was a, I mean, you think about this, like between AWS, open source communities, the CNCF and Broadcom, I mean, three really big shakeups in three areas that have been really Stable for a long time. So, yeah, absolutely. Shaky start. All right, let’s move on to Q2. Aaron Delp Well, let me, I’ll start with, I’ll take cloud in Q2 then. So Q2 starts, big first acquisition of the year. IBM announces intent to acquire HashiCorp for $6.4 billion. So HashiCorp, obviously, we cover them a lot on the show, famous for things like Terraform and a whole bunch of, you know, Vault and other, you know, kind of open source tools. They had gone public. They had begun to monetize some of it. So I think they were doing maybe $150 million in revenue or so. A lot of it, you know, if you didn’t dig into it, was Vault. But IBM, you know, intends to announce this. Just to put it in context, you know, when you make a big acquisition like this, even though HashiCorp, the company, isn’t huge, you know, probably maybe a couple of thousand people. I don’t know the exact number. We are now at the end of December and that acquisition is still not closed. So when you, you know, when you’re spending a lot of money for a big company, there’s lots of sort of regulation, lots of approvals from lots of countries that has to happen. So that was the first thing that kicked off. Second big thing, the cloud providers started putting their earning numbers out there. Remember, you know, 2022, 2023, some folks were getting a little concerned that maybe the big clouds had sort of stopped growing at the pace they were going to grow. Maybe people weren’t going to adopt the cloud. Their earnings were back to normal. In fact, you know, we’re starting to see them start to put their AI numbers in there. And a lot of that was, you know, people chewing up GPUs, not necessarily their companies, their customers making money, but they were making money. Amazon went past a hundred billion dollars in revenue. Big thing. And, you know, as Q1 bleeds into Q2, AWS said, nope, we’re tired of you guys being concerned about our strategy. We’re concerned, you know, we’re tired of you thinking we may not be knowing what we’re doing. Matt Garman gets named as the new CEO in May of 2024. So by that time, you know, the cloud providers had sort of stabilized the ship, all three of them. Google, in fact, was sort of showing sort of more breakaway than the others. But yeah, $100 billion for Amazon. Amazon had a new CEO. And, you know, we were starting to see some acquisitions come back into play, which was interesting. (Time 0:10:26)
  • LLM Market Saturation
    • Many enterprise companies, including Red Hat, IBM, Snowflake, Databricks, Observe, and Honeycomb, have entered the LLM market.
    • This raises the question of market saturation and whether there’s enough room for all these players, as highlighted in the book “Play Bigger.” Transcript: Brian Gracely And I’m really glad that NVIDIA picked them up. But then, and this is the, like, if you follow, like, there’s a book out there called Play Bigger. I encourage everyone to go read it because it really helps everyone understand marketing and categories and business categories. But lots of enterprise companies (Time 0:15:59)
  • AI Use Cases and ROI
    • Leading AI use cases in 2024 were chatbots and developer co-pilot tools.
    • Measuring ROI on these applications is challenging, leading to questions about their value. Transcript: Aaron Delp Ahead. Yeah, it does. You know, like the initial ROI is really kind of a weird thing because, you know, if you said, hey, let’s summarize, you know, let’s summarize 2024 in terms of like use cases that people Can point to, it’s going to be like, you know, aside from the, Hey, I, you know, I made a new avatar for my social media thing and I look, you know, 20 years younger and handsome and all that Kind of stuff. Um, you know, people are going to go, well, we, we, we, we deployed some chatbots and we gave our developers this, uh, this, you know, co-pilot tool, right? Like those are the two most common sort of use cases and variations of the chatbot stuff. And yeah, the tough part with the ROI stuff is I got to imagine there was a whole bunch of CIOs or even CEOs who did the thing that they did back in the cloud days where they were like, okay, I went to a conference. I talked to somebody. They said it was going to be awesome. And they started deploying it. It was like, oh crap, it’s not, it’s, you know, it’s, it’s not as much as I thought, but the hard part about this one is going to be, so like, if you’re trying to figure out ROI for like chatbots, Right? Like you start to very, very quickly get into a thing where you, you either go, okay, is 20 or 30% better? Like we’re resolving, you know, cases for my call center better or whatever, or do I have to start counting in people? Like how many people did we get rid of? And I, and I don’t know that, that they’re ready to sort of start doing that stuff yet. Right. And, and that may change over time, you know, like ebbs and flows of, of how people want to talk about stuff. (Time 0:17:55)
  • Q3 Outages and Acquisitions
    • CrowdStrike’s massive outage in Q3 affected airlines and highlighted system interconnections.
    • Google abandoned its plan to acquire Wiz, surprising many given Wiz’s $23 billion valuation. Transcript: Aaron Delp But, uh, yeah, that’s, that one’s going to be an interesting one. Okay. So we’re, we’re halfway through the year. Uh, you know, we’ve got, uh, the clouds are back to making money. Um, we sort of know where the, the, some of the stalwarts are. So Q3 comes along and Q3 is always a weird quarter because usually not a lot of announcements, at least big announcements, people are going on summer vacations. It’s, you know, people are thinking less about work and trying to think about the, you know, about, you know, getting away for a little bit. The kids are out of school. I’ll take the AI sort of starting point. This was a lot of sort of big, big picture things happening in the summertime. So first time you see the United States White House, you know, the United States sort of government making some statements about AI and open source. And this is, you know, this is where you’re starting to see this big fight starting to happen with, um, sort of Silicon Valley, the VC class, right. So the, you know, the A16Zs, the, you know, all the venture capitalists, and they’re starting to call, you know, they’re starting to say like, Hey, um, even though they’ve benefited For years and years from the growth of the internet and the growth of Facebook and the growth of, of Google and all these sort of things. Now they’re like, okay, if we start investing in AI, like our companies are going to lose to those guys, right? Cause they’ve seen, they’ve seen companies getting acquired and they very much want, um, to see, you know, open source be kind of required. They want, they want the government to sort of step in and say, Hey, you know, open source should be a requirement of AI. And the US government kind of comes back and goes like, no, we don’t necessarily want to do that. And that’s a vast, vast understatement of all the complexities of what happened. But anyways, it was the first time you started seeing the US government start to make some statements about direction they might go with AI regulation, right? You start seeing some cracks in the armor for NVIDIA, right? The Blackwell chips are delayed, at least, you know, like you said, they were announced. A lot of them have been sold out, but they’re not necessarily coming out as fast. So there’s a little bit of concern there, right? And that concern about NVIDIA has been that we will see this ongoing theme throughout the year, even though their stock continued to, to kind of, you know, break all sorts of records And so forth. Um, you keep seeing open AI working on this huge round of funding. Um, you know, and, and I say huge round, like huge in the context, and we’ll get to it towards the end. Um, not so much the amount of money, but the value of the company, right? So, you know, trying to make sure everybody was going to do that. And this is where you start to get into sort of some interesting things where, you know, Apple was going to invest, NVIDIA was going to invest, you know, like it was like, okay, who are The investors and do the investors look like they’re going to become the right partners for open AI or not? You get towards the latter part of the quarter and the government stuff starts to kick in again, right? And you covered this on at least one of the shows. State of California had their own AI safety bill, which got rejected. And this caused all sorts of turmoil with VCs. It was kind of a complicated thing. The EU started having their own sort of rules around this. And then you started seeing some numbers fly out there, right? So OpenAI said, hey, we’re losing 2 billion plus a year, right? So we’re making $3 billion, but we’re losing $5 to $6 billion a year. And this is where you start to see the very much change of OpenAI from, you know, we are this nonprofit to they’ve officially filed to be a for-profit company. And, you know, we see some shakeup there. But lots of sort of big things that, that have to do with regulation that have to do with funding. You know, this is, this is like think piece central for, for the AI in Q3, because you know, there wasn’t necessarily tons and tons of stuff, but the stuff that did come out, people were Like, okay, this has world reaching ramifications and you know, lots of, lots of opinions floating around. Yeah. Brian Gracely And I would add to like one of the big things around that, cause what you see in Q3 is, is reaction to a lot of the trends in Q1, right? So like to take it back out for a second, like for folks that are listening, like, like always start to think about this of like, Hey, what does this mean? And, and, you know, two quarters from now, three quarters from now, because it usually takes that long for the ripples to expand out and action to be taken. And when it comes to the regulation stuff, I think a lot of the reasons why a lot of the regulation stuff didn’t either didn’t pass or like got neutered or got watered down or all these other Things is just because I don’t think a lot of folks truly understand how to do it, because it’s like, you know, do you regulate the model? Do you regulate the weights and it’s used? Do you regulate the training data? Do you look at all of those things? Like it just gets really complex really quick and your average, you know, politician and regulator isn’t going to be able to wrap their head around that. And so it’s no surprise the first batch of regulation and legislation fell a little flat. I mean, it’s like a first draft of anything. Aaron Delp Well, and it’s, you know, and it’s the trick of, you know, do you, you know, like, for example, like, when do you regulate, right? Like, are we at the beginning of growth? Do we want to slow down growth at the beginning? Do we want to let the thing get out away from us? Can you kind of the old thing of like, once the horse is out of the barn, can you get it back in the barn? But the other part of all this stuff is, for all the complaining about it, in essence, unless the regulation were to shut you down, most of at least the big companies have figured out, Like, I can just pay fines. I can, you know, I can, I can pay my way out of any, any rules that we break. And so, you know, I think there is a lot of skepticism about like, will any of this stuff ever come around? And, you know, the, the, the flip side of all of it is like, well, you know, the market sort of works itself out. Although, you know, the market does sort of work itself out into, you know, two or three very, very, very large companies. So understand where the – understand at least where the investment community is concerned about (Time 0:19:21)
  • Startup Advice
    • When considering joining a startup, prioritize the ability to sell options on the secondary market.
    • This provides financial security since IPO markets are not always accessible. Transcript: Brian Gracely Me say this real quick too. Um, maybe we’ll get back to doing some, uh, Sunday’s perspectives on career advice in the new year. And Hey, there’s always that, you know, open false promise that Aaron will do a Sunday perspective at some point, but I will say this and I, I won’t say which company cause I don’t want To dig too much into the past, but I will simply say based off of doing a bunch of startups in the past, if you’re out there and considering a startup, always, always, always, always look For can you sell your options on the secondary market? Yeah, yeah. Quick bit of career advice to throw in the (Time 0:29:45)
  • Return-to-Office and Cloud Growth
    • Several major companies, including AWS, Dell, and AT&T, mandated a return to the office.
    • The cloud market continued to grow, with the big three cloud providers generating significant revenue. Transcript: Brian Gracely Absolutely. And we’ve talked about it many times on the show and maybe it’s just because it’s near and dear to our hearts. But AWS, Amazon, they went back to five days, RTO. So those return to the office mandates continue. We saw Dell earlier this year. One of my former companies also announced a big one here recently. Um, I don’t know if they’re public or not, so I’m not going to say it. I think AT&T just announced a big one. AT&T just announced it. So the pendulum has swung back hard. Um, and it’s fine. Like, you know, I, I think that was just a natural progression of the market and it’ll settle into something in 2026. But anyway, um, then the big three clouds, I mean, continue to grow revenue, continue to print money. Um, the FTC and the, the DOJ, the department of justice there, they went after both Microsoft and Google for antitrust stuff. So we’re starting to see a little bit of government kind of going, hey, how big are y’all? And do you need to be that big? Right? The cloud opportunity was targeted for $1 trillion with a T. And at least that’s the first time I had seen it, you know, estimated at that. And it makes sense, you know, if AWS is by themselves doing a hundred billion, like it makes sense to move the bar up a little bit. Um, and then AWS reinvent, I think there’s, and I think this is going to be all companies going forward, but I think they’re the first ones and probably the biggest ones. Cause I think at this point, their keynotes are probably the most scrutinized in our industry. Trying to find this balance between cloud and AI going forward. Like what’s going to make money? What’s aspirational? What’s in the future? What’s on the truck today? There’s all these different triggers and levers to pull. You want to be seen as somebody who’s going into the future. But let’s be honest. And we’ve said it many times on this show, the stuff that makes money is often the non-sexy stuff. Yep. (Time 0:32:25)
  • 2025 Predictions
    • Aaron predicts 2025 will be the year of AI ROI concerns, worsened by agentic AI.
    • Brian foresees government vs. AI conflicts, especially with the new US administration, the EU, and China. Transcript: Aaron Delp Yep. All right. Real quick, let’s go through predictions, predictions that are, you know, sure to be at least slightly wrong, if not potentially completely wrong. Brian Gracely But once you go first, let’s kind of bounce back and forth real quick. Yeah, absolutely. So I’m going to say, most of mine are AI related, by the way. And I mentioned this earlier on the show, AI, ROI. I think the ROI on AI, and by the way, I talk to customers about this. And so I have these conversations on a, on a semi-regular basis. There’s two big things that is, is holding the industry back. Supply chain. So, and you know, we talked about it in videos and then ROI. Like if I’m going to do this, I’m not doing most AI projects on a three to five year ROI anymore. Like it’s on a 12 month, maybe 18 month ROI. And Oh, by the way, the stuff also costs more. So now you have something that costs more and has less time to get a payback on. And I think it’s only going to get worse. Yeah. Aaron Delp And they’re probably, I mean, this is one of these things where you think about it and you go like, okay, this is the first time since basically the mainframe that we’ve had marginal costs. You have to think about marginal cost with, you know, with computing. And but yet, you know, we used to have, you know, 36 month ROI, 48 month ROI or whatever. Like it almost seems like insanity to even put a one year ROI on it when you go, OK, it’s really hard to get chips. I can’t get chips the way I could get CPUs where you get them really short. And oh, by the way, I have shortage of people that actually know how to do this stuff. But, you know, let’s, you know, but, you know, let’s, let’s throw a 12 month ROI on there and kill the project. So, I mean, I think you’re going to see a lot of people claim that they’re trying to do that stuff. And you’re going to see a lot of articles that are like, Hey, CIOs realize that 90% of projects fail, you know, AI projects fail because, you know, they gave up on them or they were too expensive Or whatever they are. So, yeah, that’ll be interesting. First one I’ve got is I think, and this isn’t so much a prediction, but just like a thing I think we’re going to see throughout the year is sort of government versus AI. So whether it’s… It’s, you know, the new administration coming in with in the US, whether it’s the EU starting to, you know, do their thing where they’re trying to get out in front of it and turn it into Another, you know, the EU Act and so forth. You got, you know, people doing economic concerns about what’s going on with China. China doesn’t necessarily want to regulate their stuff, but like how much are things regulated, you know, sanctions against China and some of those things. I think there’s going to be a lot of sort of government versus AI going on. And unfortunately, I think it’s going to be very, very complicated to sort out, you know, how much of this is government. Like what is the regulation trying to accomplish? What do we think the byproduct of that’s going to accomplish? And, you know, does this seem like a short-term fix or a long-term fix to stuff? But I think that’s going to be a big theme throughout 2025. Brian Gracely Yep, absolutely. I have another one. I’m kind of a Debbie Downer for AI this year, considering I kind of do it as my day job. But here’s another one. But I do have a positive one on my third one. But you’re hearing more and more about agentic AI. So and for those that aren’t familiar with it, basically think of it as instead of it generating things or predicting things, it actually does things, right? So it’s an agent. There’ll be at least one, I’ll call it Skynet scare. And the reason, what made me think about this is there’s an AI podcast I listen to. It’s two guys. I think they’re out of Australia. They just love to just play with this stuff and see what kind of things, the crazy things they can do with it. And for whatever reason, they love to set it up for gambling. Um, and, and so they, they had a set up two systems. One of them was trading penny stocks. Um, and the other one was doing online poker, like doing vision, analyze the cards and do whatever. And, oh, by the way, both of them made a ton of money, which violated any and all terms of service. And they knew that and they were just like, hey, we just want to play around and see what happens. But they actually shut everything down because it was like, man, we’re starting to make a shit ton of money and we’re going to get shut down here before too much longer. And I think it’s only a matter of time before something like that happens where somebody’s just playing around with something or let’s be honest. I mean, again, you take the stock market. I’m sure all of the big fintechs out there are developing some kind of agentic AI system that’ll be able to trade a millisecond faster than everyone else. And who knows what that’ll do to the stock markets. So I think there’s going to be at least one big national, if not world news, Skynet scare. (Time 0:34:32)
  • Agentic AI and Gambling
    • An AI podcast experimented with agentic AI for gambling, leading to significant profits and terms-of-service violations.
    • This highlights the potential risks and ethical concerns of agentic AI. Transcript: Aaron Delp Yeah. Now, it’ll be interesting. I saw sort of a flip side story where somebody had done that. They had run some stuff and they did the thing where it was like, oh, it ordered pizza by itself. But at the same time, it got into some sort of auction trading. Auction, I don’t know the right word is, but trying to bid on projects for something. Think of Fiverr or something like that. And basically, the system, because it was so deterministic to get to the end result at whatever was considered best, it basically ended up bidding on the project at zero cost, you know, Like, Oh, we’ll do it for free. And so I will win. And the complete side, you know, the complete opposite of your making tons of money was, you know, I, uh, I raced myself to zero because I could do it 24 hours a day, three, three 65. So yeah, there will be, there will be several of those. And, and, you know, there will be the, the, the stories that will come out of it of like, Oh, we can’t trust this stuff. So yeah, expect that. Um, I am going to, uh, for weekend perspective, you mentioned weekend perspectives. Um, one of the things that I’ve been kind of beating the drum on a little bit is just this, this search for, you know, interesting AI applications. And I know there’s, there are tons of people you can just follow on Twitter and blue sky and others that, you know, are doing nothing but kind of like, you know, kind of going, hey, here’s 25 new interesting things. I am going to try and spend some time each week looking for stuff that, you know, from like an enterprise day-to job perspective, like is truly sort of interesting. And I think I’m going to end up doing some weekend perspectives on those as well. So I’m sort of out looking for stuff outside of your typical sort of chatbot plus plus or developer co-pilot stuff and some things like that. So we’ll be doing some more of those from a weekend perspective perspective. You know, I don’t necessarily be like going out and interviewing the companies like we typically do, but at least looking for them and figuring out like, OK, is this an interesting pattern Or is this just, again, you know, some really tiny little shim layer on top of a chat bot or something like that? (Time 0:39:11)
  • Inference and Pricing
    • Aaron predicts 2025 will be the year of AI inference, where companies focus on using existing models rather than building new ones.
    • Brian highlights the need for AI pricing to evolve to bridge the gap between experimentation and regular use. Transcript: Aaron Delp Nice. Nice. Brian Gracely So this is my positive AI prediction for 2025. I think it’s going to be the year of inference. And what I mean by that is like, think about it. There’s always been two big training, tuning, fine tuning, getting it ready for production, and then inference actually putting it into production. We always talk about training and fine tuning and how expensive it is to build these models. And I mean, at the end of the day, that’s a non-profit loss center for the business. Well, inference is where things make money. And so I think we’re going to see a lot of companies kind of make that connection and go, you know what? I don’t need to build stuff. I don’t need to customize it all that much. I can just take a model, throw some rag on there and get it in production with inference as quick as possible. Yeah. And I think, you know, inference is going to be like, that’s like, you were talking about where they, you know, the big applications and all that. But I think the stages also matter. And so the sooner people can get to inference, which means, you know, it’s a money making or optimizing or doing something else that’s on the plus side for the business, that connection Will be made this year. All right. Aaron Delp So you’re predicting, you’re predicting positive, positive revenue for, for a lot of these applications. Interesting. Yeah. I think that’s going to be, you know, I know Amazon was talking about that. I mean, AWS is talking, trying to talk about that a lot in, in reinvent. It’ll be interesting to see if, if they get to that point, you know, beyond, beyond a few them. Yeah, I think the same thing, I think it’s going to be for me, you know, it’s not so much a prediction, it’s just a thing to keep an eye on is like, what does the pricing begin to look like? Because I don’t know that people necessarily have a good sense of like what some of these things should cost, right? I think, you know, when Amazon was first being like, well, this, this size CPU cost eight cents an hour, you know, even that was sort of complicated. Um, but, uh, I’m going to be very interested to, to watch. I’ve been playing around with a whole bunch of things and like, it feels like the, the gap between like, let me try this stuff out and like use it on a regular basis feels like big steps in A lot of cases. And so I’m going to be very interested to see if we start to see the gap between like, you know, get people interested in this stuff and the first level that they can start to do some things, Or, you know, do we still have these big jumps? And I think those are going to be, you know, kind of directly correlated to the thing you said, which is, you know, if, if inferencing costs can’t come way down, right. And they probably need to come down what 10 X from where they are today. Yeah. You know, just, you know, if, if it’s the normal sort of, um, you know, curves that we tend to see, um, you know, I think we’re going to see a lot more people who, who get stuck in that first Sort of tier. Like I, well, I did free stuff. I kind of did some things, but I wasn’t, I wasn’t seeing enough, you know, value versus cost, you know, to, to move to that next level. So I’m going to be trying, I got to figure out a way to sort of track that because the last thing I want to do is have a hundred, you know, random AI accounts kind of dangling out there that I’m paying eight bucks a month for 20 bucks a month or whatever. Brian Gracely Cause I’ll forget about them, but I, I am. It’s the, it’s the equivalent of us having like 10 streaming plans. Yeah, exactly. Well, yeah, exactly. Aaron Delp That’s exactly what it is. So yeah, kind of watching that, you know, individual value versus, you know, versus sort of the dragging on of, you know, sort of stepped costs to get to these sort of things. So and that’s, you know, part of that’s where you see people, you know, going like, oh, we have, you know, 80,000 developers, you know, running copilot. And you’re like, yeah, that’s because you didn’t have to think you just bought a, you bought an SLA. And, and, but, but for like individuals, I still think it feels very clunky. It doesn’t feel built into anything we do necessarily yet. And so I’m keeping an eye on that stuff. So, right, man. (Time 0:41:14)