If You Own AI Stocks.. GET READY FOR A BIG CHANGE!

Hey everyone and welcome back to another video. For today in today's video we have a lot to discuss. We have snowflakes earnings report outcome huge announcements from Meta and Google. Some comments from OpenAI Sam Oldman on the use of water in data center and how they should maybe tackle this whole misinformation problem around AI. We're of course going to start with a big overview of what's going on in the market right now because yes, over the last couple of days it's been a bit shaky. We still have some macro data that should come out. Could influence of course the Fed's decision to increase rates. Leave it flat. I don't think they will cut anytime soon. But right now, the market is expecting rates to increase. So any data that could support a pause still a continued pause in rates is definitely going to be very wellreceived in the stock market because right now yes it seems like everybody is convinced that rates are about to increase. By the way inflation not being under control in the United States well it's not just in the United States. In various countries in Europe inflation is also picking up a little bit. Now, if we go and have a look at the stock market, this is how we closed out yesterday. Yesterday being on a Tuesday, it was okay. I think we started off way worse. Ended up with Nvidia being up 3.2%. Meta was up 2.4%. Netflix was up 2.3. Micron is up 2.4. So, all in all, big names were up. Palanteer was actually down close to 6%, PaloAlto was down 9.28% to 8% despite them reporting quite a good earnings report which means that yes the other cyber security names are also getting hit as you can see crowd strike was down 5.4% 4%. Quickly looking at the Neocloud players, Irene was up 7.5%, still under $40 per share, we have Cororeweave, or as some people call it as of right now. Poore. It's a very, very good name. Whoever came up with that, congrats. Cororeweave was at one point back under $80 per share. We're now just over it. So, we'll see what happens with that. It's still a very, very much hated name in this industry. Nebus did go slightly under $200 per share. Closed out the day at $24. As for the big tech companies, as you can see, most of them were green and whoever wasn't green was just slightly slightly red. Semiconductor names here mixed back. But then again, whoever is red is slightly red and most of who were green were also slightly green other than Micron and Nvidia. Surprisingly enough, fintech names actually got a lot of love on Tuesday. The only name here on our list that was red was Weeble. All the rest were green. They was up close to 10%. Shift for 7.6. New was up 6.5%. Settle also 6%. Le SoFi that was up 4.6, but we are back under $18 per share as of right now. PayPal is still still very easily in the mid $50. So that was quite surprising to see yesterday. Now before the market opens here as well, it's a bit of a mixed bag all in all, but not that surprising. The biggest moves here are just iron being down 2% and Snowflake being up 24.1%, Broadcom is down slightly, I believe around 3.2% or so. We're going to cover everything you need to know in this video. Now, as you can see, Broadcom right now is up 17% over the past year, and year to date, it's only up 1.8%. Of course, if we zoom out a little bit more, tremendous record here, 305% increase. It's a 1.68 trillion company, trailing PE 45.2 times, forward one 22.4 four times. And here as well, just like with an Nvidia, the price to free cash flow is of course much lower than all of the other big names in the AI space because yeah, they are the one making a ton of money. If we look at the revenue growth for the quarter, they reported 29.5 billion in revenue and then increase by 85.5%. If we look at what's expected over the coming three quarters, you can see the following thing. Upcoming quarter expected to increase and accelerate 94.2% then 92.4 and 83% year-over-year growth to reach 40.6 billion dollars. Now this is clearly a company with still a lot of momentum. And so if we look at the major overview for this quarter, the big number here of course comes from the fiscal year 26 AI semiconductor revenue 21.7 billion up 236% year-over-year. $230 billion. That's AI revenue guide for fiscal 28. The AI semiconductor revenue did beat guidance by 4.4%. Infrastructure software missed by 1.6%. They partner with Apollo and Blackstone on an XPU focused financing vehicle to fund over 20 gawatt of compute for OpenAI Entropic through fiscal 28. That structure includes up to $29 billion in residual value guarantees, the main source of the investors criticism. Gross profit margin guided down to 73% from 78% as a mix and memory cost rise. A bit margin still guided flat at 66%. Now they talked a little bit about the customer ramp. So we have Alphabet here. The TPU version 8i in for inference. The shipping here matches or beats Vera Rubin on infrance openai that's the jalapeno a chip they're also expecting here 1.3 gawatt in 26 to go to over 5 gaw by 2028 as for entropic a TPU there coronwood 1 gawatt in 26 largest xpu customer by 2027 and as for meta mtia 3 gawatt total capacity through 2028 and then comes the I Maybe surprising, maybe not surprising comments from management. Octan said that a cheap code designed around a customer specific LLM workloads will outperform any generalpurpose GPU on that specific workloads. Basically calling out here Nvidia or maybe even an AMD although an AMD is a bit different here. Then the other comment I want to touch on is the comment that well they were asked about the Alphabet partnership because well Alphabet has invested in MediaTek they're working with MediaTek and so they said the following thing they attribute the durability of the Google TPU relationship to Broadcom's IP lead to chip interconnect HBM SRAMM integration and advanced packaging. Guess the market is always fearful. We've heard that from Marvel I think a year or so ago where the market was well fearful that Marll could lose Amazon as a potential customer. You know what happened since then? Marvel went from 70 bucks per share to much much higher uh than that and became an even more important player in this whole industry. And so you might say, okay, if everything was great, despite his margins maybe coming down and things like that, why is the stock down a little bit? Well, again, it's more about market maybe asking even more questions first. Thinking about 27, 28, yes, maybe concentration, risk, entropic, open AI. What if they cannot grow as fast as expected? How does that impact a business like a Broadcom? And let's say let's say they're correct here that a chip code designed around a customer specific LLM workloads will outperform any let's call it Nvidia GPU. Well Nvidia yes right now 21% I think two quarters ago 21% was one customer 17 another 16 another and then we've got now a new customer probably SpaceX AI around 10% of revenue but the ecosystem that they have is just multiples time bigger than a Broadcom. So yes, if let's say an entropic does slow down, even though for Nvidia I think entropic is much lower risk than for a lot of the other companies out there. If let's say Antropic slows down, OpenAI slows down could happen and I'll touch on exactly why in a bit. I think it hurts a company like Broadcom way more than it hurts a company like Nvidia. Now to understand why entropic and open eye might I'm not saying they will I'm saying they might experience a slowdown purely look at the most bearish case here is because of what's been announced here and yes of course another day another video another meta topic although in this time we're also going to include Google here so Muspark 1.3 came out it's rolling out in Muse code and meta model API and remember what we've said once Meta came out with Muse Spark the baseline has been said they are going to move much much faster than before because they feel more comfortable building upon this base and that's exactly that what we're getting here. What's new? Improve performance across agentic encoding task building on months of real world use of Muse code and meta model API positioned as a step forward metastated goal of personal super intelligence and previously available reasoning modes shipped today. A max reasoning mode is coming shortly pending additional safety testing. Now, what's the most impressive here is this is the price per million token. It is dirt cheap. Okay, the contributor tier is priced at roughly 12 times cheaper because usage data feeds back into Metas model improvement pipeline. The standard tier is not used to improve Metaas product. Now what's important to understand here is this is still very very early days. Okay. To really cause any problem to open AAI an entropic or anybody else, the adoption needs to go up. The ecosystem, the developer ecosystem needs to grow. They need to ship. They need to improve the models. Everything needs to improve because yes, this is already great. This is very cheap performance-wise. Incredible. But it is early days. Now, let's say we fast forward a couple of months down the line. watermelon model comes out, they ship it out, it's available to everyone, maybe everything gets integrated already in WhatsApp, Instagram by then and usage goes up. Well, that's a completely different story because if suddenly your developer ecosystem, your usage goes up significantly, the output for me, the end user, remains the same or it's even better and much cheaper. Well, why do I have to spend so much more money on the other models out there? Right? Meta, the whole point that I've been saying for the last couple of months here is Meta's core business is extremely profitable. This is something that we've seen with a Google as well. So, yes, they can, let's say, take their time. Although with Meta, they took too much time in my opinion, but still they're now coming out with we could call this state-of-the-art models. They can come out with these types of models. could be the same performance as a Fable 5.1 or Soul or whatever, but they are allowed to use pricing power because their core business is so profitable. The others, if they want to keep up, reduce the price all you want, but if you're planning to go public, you have to show that well revenue is going up, ARR figure is going up. Well, if you reduce your pricing so much to try and compete with a giant like Meta, I don't think you can win. But as I said before, it's still very early days, but this is already a very good sign. Then we're going to Google. They released Gemini 3.8 Flash and 3.8 Flash Cyber. Fortunately for me, I'm still stuck on 3.6. Maybe it's because I'm in Europe, but I've seen people I think in Canada that are also stuck still on 3.6. So, what the heck's going on there? I don't know. But Gemini, Google is another case of yeah, they might not be releasing or they might not be at the frontier anymore. But costwise, they are still best-in-class. They are of course focused more on growing Google Cloud because Google Cloud is extremely profitable. They are of course focused on making sure that yes, Gemini for free, AI overviews, AI mode, all of that, all of the free features are growing. Multi-active users reach a billion. So they don't care. They give away free stuff that works, that's helpful for the majority of people. And luckily for them, they have a whole other business that is being extremely well monetized. That's also how and why they will keep on winning. So 3.8 Flash, same intro price as 3.7 Flash. They've seen gains across software engineering, agentic task, and multi-step reasoning. As for the 3.8 Flash Cyber, this is a defender model. Right now, this is only available to trusted defenders via the new fair wind program and I believe government agencies. But Google, as for the reports that have been coming out, it is believed that Gemini 4 is actually going to be an extremely good model. Right? Since the release of Gemini 3, Nano Banana, all of the hype, everything's been quite good, but it has not been as good as what I guess the majority of people wanted it to be. But it is believed that Gemini 4 could be the next huge release, huge improvements across the board. I love my Google position. I think it's a great company, the best ones out there. But yes, if you look at what you are capable of doing with cloud or open AAI and then you go into Gemini, there's still such such a big gap. Moving on to Snowflake's earnings report. Snowflake has been doing extremely extremely well over the past 12 months. Revenue right now for the quarter was up 36% year-over-year, $1.55 billion. Net revenue retention at 126%. New fiscal year 2027 growth guide sits at 36% that was 29% before so excellent there. EIT margin is up 400 basis points yearover-year. The beat revenue estimate by 5% product revenue beat guide by 5.1%. Annual revenue guidance was raised by $230 million well above the $70 million Q2 beat. So significant confidence in the second half. Growth is organic. No single large model partner deal is propping up the numbers. AI agents coding and coworker guru accounts from 7.1,000 to 9.1,000 and 5.2 to 5.8,000 in one quarter. It is definitely not a cheap name, but it is one that is growing significantly faster than before. As for why Snowflake ties its agents Coco and Co-work directly to its data platform, the framing of the agent quality was this outputs are only as good as the data business context and governance an agent can actually reason from. We've of course heard this same thing from all the other companies that are doing quite well especially in software and with the AI adoption. So with Snowflake again quality company growing fast acceleration becoming better and better more important definitely not cheap but they have a lot of momentum. Lastly here before showing you that clip from Sam Alman and Scale apparently they are going public sooner rather than later. They're talking $103 billion in contracted revenue. That's what they said. That's up from 51 billion before the entropic deal. The entropic deal is $45 billion. The average contract length is around 5.7 years, which is around $18 billion per year. The actual quarterly revenue is still extremely small, but growing very quickly. Q126 was $37 million. Q2 was $100 million. Of course, as more and more of these big AI private companies go public, whether it's infrastructure, energy, or AI models, the more information we get as investors, the more information, the better. Now, talking about information, or shall I say misinformation, water usage, AI data centers, etc. Let's have a listen to what Samman had to say. And I know whatever he's saying, maybe take it with a grain of salt, reduce it by 20%, 30%, do whatever you want, but still listen. >> It was like every time you run a single CHBT query, it's like, you know, you ran your shower for like 6 hours and the water never comes back and it's just it's done. I don't have the exact calculation in front of me, but I I think the real number is something like doing this from memory, it might be wrong, but it's close. For every 38,000 chatbt queries, um that is the same amount of water that is used in the production of a single almond in California, which is like really and this is like the full on, you know, total true water accounting, not just what's running in one data center. There's a question of like where this came from because the people that are scarfing down 12 almonds at a time don't feel like they're doing something horrible from a water perspective for the most part. Um it is [clears throat] true that data centers at one point used evaporative cooling. Um but they have not done that in a long time. Like if you look at a modern very large data center, it uses the like equivalent amount of water as an office building in terms of you know people like running the sinks and the toilets and whatever. Um so that has been a um robust meme and difficult to disprove but I don't think holds up to any scrutiny. >> No, I know I know what people might say. Of course, he's going to say this because well, if there is more negative sentiment towards AI and data centers, it hurts his business. Correct. But I guess there are many people that can actually go there and check check how much water they're using at OpenAI at Tropics Data Center, Google, Amazon, Nebu, Cororeweave, etc. You can go and check you can go and check that then go check the almond story. Of course, if it comes out to be true, what would be the headlines? And you know what I do think that they are becoming extremely extremely efficient the data centers with water use with everything else. Now then comes the perception of it. How do you change this from a public standpoint? Well more of these CEOs more of these big AI leaders need to come out not just talk not just say what he just said but come out with actual proof. Go to a farm check what's happening there. Go to your data center, show us what's happening there. Go on TV, go on social media, expose it all. That's how you change perception. Now, I know what people say. Yeah, but do you want a data center in your backyard? Please show me. Please show me 10 people, 10 people that have a data center in their backyard. You cannot. Okay? There are no data centers in someone's backyard as of right now. And if there are certain permits that allow companies to build data centers extremely close to communities, well, that means that whoever is in charge of that community is an absolute idiot because they allowed these companies to build data centers near the community. I've been to the United States multiple times. Every time you take the plane, you get to look out of the window. you have enough space for hundreds of data centers that are far far away from any community. Okay, so space is not really an issue. I know I can make jokes about data centers in space. No, but space right now on the ground is not really an issue. I don't have a nuclear plant in my backyard, right? I think Joseph Carson make a very good point on X. You don't have huge factories, chemical things, all of that in your backyard. No, you don't have them. You have them in specific areas in industrial zones. Same thing here with data center. But of course, when there is a very very popular campaign going on on social media, on Tik Tok that's been influenced by outside powers or just by idiots, then yes, then everybody's going to believe that this is true. But if we're being honest here, it is not. Now, of course, it's going to be a long long way ahead to try and now change the whole narrative. Hopefully, they'll do a good job, but I do think that this whole misinformation stuff will continue just not just for AI, but for plenty of other things. If you see an adversary, a competitor moving quickly, it's quite easy. It's quite easy today to manipulate the progress to make sure that the progress slows down. plenty of useful idiots out there these days. Now, as for the whole AI story, the whole AI story becomes even better and more interesting right now because as you've seen, models come out, they become way more efficient, way cheaper. So, go to go back to the point that I made in the previous video. We need a hell of a lot more compute. Yes, we do need more compute, but as I said, the GPUs are becoming more efficient, the models are becoming more efficient, and they're becoming cheaper and cheaper. Who wins? Well, who wins? The cloud players, the big AI companies, software companies, they're all going to win big because they can all move faster, cheaper, more efficiently. If your business is only selling subscriptions to get access to your LLM, you've got a problem. And as I said before, AI is not going to slow down whether entropic goes away, open AI goes away, or another player goes away. AI is going to continue. This revolution is going to continue. The compute, the demand, everything is just going to flow from one player to the other. That's about it for me in today's video. See you all in the next one. [music] Bye-bye. >> [music] [music]