저자: Kevin · 외국인
요약 보기
OpenAI 2차 시장 매도 난항과 과도한 밸류에이션(약 35배 매출) 지적, Anthropic 상대적 선호. 메모리 가격은 급등 후 조정 중이며 마이크론·샌디스크 PEG 기준 고평가로 탑 신호 언급. 사모신용 대출은 안정화 추세라 평가. 구글은 PEG 1.6으로 저평가로 봄, AMD 저평가 언급, 엔비디아는 탑 가능성 언급.
다듬은 글
We need to catch up on what's going on with private credit because private credit is starting to sound a whole lot like the excuses we've heard in 2007 and 8. Open AI shares are apparently now getting dumped in the secondary market. Sometimes people apparently can't even find buyers for the open AI shares.
At the same time open AI is bragging about having just raised even more money on a primary offering. Google just stabbed memory chip producers in the back. But memory chip stocks are already rebounding. So, what's going on and how do these all relate together? Well, in this video we're going to break it all down with our friend the bum mom hat.
All right, let's get started. So, let's just catch you up on open AI. So, you may have already heard that open AI has begun offering private equity firms a guaranteed minimum rate of return of 17.5%. Now, typically when companies offer a guarantee, it's a really ballsy move. That's like you guys are extremely confident.
But when you have to offer 17 and a half percent as a guaranteed minimum return, it starts raising the eyebrows of maybe you guys are just raising at too high of a valuation. And that's likely exactly what's happening, which is one of the reasons people seem to be flocking towards what? Anthropic. Take a look at the valuation difference between the companies and you'll start understanding a little bit of what's going on here.
Anthropic is raising money at about 380 billion dollars. Whereas open AI is now sitting post money somewhere between 830 to 852 billion dollars. So, you're getting Anthropic for about half the price. And frankly a lot of people like what Claude is doing, especially with Claude code and I mean even after the front end leak of their data set, people were kind of excited about Anthropic.
They like the models. And Anthropic has done really great work. I personally find that one of the things that's happening is it's sort of like this this horse race where it's sort of like now open AI is in the lead and then all of a sudden Gemini is in the lead and oh open AI is trying TO COME BACK.
OH, AND THERE'S ANTHROPIC. YOU KNOW, IT'S it's literally like you're at a horse race where they're all kind of going to the same commoditized destination where these models will all be pretty dang good where it's sort of like all right, like is it really that much of a difference between driving that model and driving the other model?
Probably not. And so, that's where we're heading, but what's going on with this struggle to potentially sell open AI shares? So, where this talk is coming from is that private equity firms are being promised these 17 and a half percent returns. They're being promised early access to open AI models. And on their primary offerings they just secured commitments for fundraising another 122 billion dollars.
This is inclusive of that last round we heard about a month ago. This is just sort of an extension of that same round where Amazon's putting in 15 billion and then 35 billion if they go public or they get to AGI. We're also seeing Nvidia and SoftBank commit 30 billion dollars. It's a lot of money for a company that does about 2 billion dollars a month in revenue.
Because if you think about it, 2 * 12 we're sitting at about 24 billion dollars ARR and they're raising at you know, 852 billion dollars now divided by 24. Puts you at about 35 times sales. That's not profit. That's 35 times the top line sales figure. And the scary thing about that is we still don't think open AI is actually profitable.
Which means you don't even I mean you can't even calculate a price to earnings ratio because there are no earnings. We've also had a lot of talk historically about potentially subscribership starting to slow down at open AI that the easy gains have been had for chat GPT and people have moved to Claude or Gemini or whatever.
So, as a result of all of this, there's talk that shares have become almost impossible to offload in the secondary market. And to me, when you can't sell something, it's always a sign of one thing and it's called price. But literally Bloomberg has this article about how there are companies looking to sell 600 million dollars of shares of open AI.
And they're seeing a drop in demand for open AI with now half a dozen, so sixish hedge funds and venture capital institutions interviewed suggesting they cannot find buyers for open AI. They're literally saying, we literally couldn't find anyone in our pool of hundreds of institutional investors to take these shares.
Buyers have indicated they have money ready to deploy for Anthropic though. Right, because the price is better by a factor of two essentially, more than that. And people are flocking to Anthropic because it's the hot tamale today. That's unfortunately one of the downsides of private credit is because you don't have a public market that immediately values these shares.
You know, you would actually if you did, you would actually see open AI share price fall and Anthropic's rise. Instead, when you're in private equity you're sort of gated with who's in your Rolodex and who you can call. If you can get allocations for Anthropic and you know, then then people you know, try to get those allocations to Anthropic at lower valuations through banking connections or VC connections or whatever.
And then they don't buy open AI at the crazy valuation. And nothing in the market tells open AI other than this obviously, but publicly that hey, you know, maybe your price is just too damn high. Maybe you guys aren't worth almost a one trillion dollars. And this is obviously coming at the same time as they're trying to explore revenue avenues.
Now they're saying they're not going to get into exotic content, so we're not going to get a revenue source from there, which is probably a good thing anyway. Not sure the mental health implications of that. And on top of that, open AI has now been testing ads with about 20% of their users who are eligible for them, which is about the 80% of free users they have on the platform.
And they say they're at 100 million dollars of ARR, which is great. But remember 100 million dollars of ARR divided by the 24 billion dollars that they're making off of subscriptions, 100 divided by 24 billion basically is you know, less than half of a percent of their revenues are coming from ads right now.
Now, it's good that they're getting some revenues from here. We don't know what the costs are for them to integrate these ads. I can't imagine too terribly much depending on what kind of sales teams they have for this. So, all to be determined. But we can see open AI, you know, they're starting to really show some funding stress, especially when Amazon says, look, we're not investing anymore.
Nvidia is saying we're not investing anymore unless you guys go public. The pressure is on to go public and actually get realistic prices on these companies. And reality is as these commoditized they'll probably all compress. Now, something that has been compressing recently has been memory prices. There have really since about last Saturday there have been a lot of reports about memory prices sharply falling off.
You have to know this about memory prices. Memory like I always think it's so funny when people are like, oh my gosh, memory prices are collapsing. Dude, memory prices are sky freaking high. We did an analysis in our course member group on Sandisk you know, like six or seven months ago. Yeah, six or seven six or seven months ago and we're like, wow, you know, this this these numbers are fantastic and I'll show you some of the financials in just a moment.
But the numbers at companies like Sandisk are fantastic. I do think the stocks have these levels topped out. I think they're they're very rich now, but you know, they're like two and a half times as expensive as they were back when we did our analysis previously on these. But anyway, people are suggesting that memory prices are in total collapse, that they're just bombing out.
But you have to remember memory prices just shot up by a factor of about seven and a half times, which is massive. And they've come down by about 20% from the peak, which is a fraction of the movement that they've seen to the upside. Part of this is because of Google's Turbo Quant research, which suggested that Google researchers have found a way to reduce the memory needs for large language models.
You know, when you go have a conversation with AI, you want AI to remember the context of what you're talking about, you know, 10 messages later. It used to be back in like 2023 and 2024 that it would start hallucinating and forget as the conversation went on. Now AI has gotten pretty decent where it can actually still contextualize some of your early messages.
That is stored heavily in what's called a key value cache inside of the large language models. And that is one of the big places that we need a lot of that high bandwidth memory. Turbo Quant says that we can actually reduce that memory use by a factor of 6x. Some critics say that's totally overblown.
They're not going to get to a 6x reduction. But that definitely didn't help memory prices. On top of that, you've got helium concerns because of the lack of LNG exports out of the Middle East. One of the byproducts of LNG liquefied natural gas production is helium. So, helium prices are rising, so input costs are going up, which could potentially damage memory margins.
Price growth is slowing, right? There's a when when you're no longer going parabolic on the price of memory, you're going to eventually see that that show up in margins. And we're actually going to go look at an income statement here for Sandisk in just a moment. Of course, there's panic around the Middle East and CapEx concerns, Chinese memory competition.
All these things are playing into sort of a topping out in the memory sector. Just totally reasonable, mind you. But some like a little side tip here and then more on this memory cycle. Side tip, when you're talking to AI, I highly encourage you open many new context windows and even use temporary chat because you don't want an AI to bias towards what you've been saying, what you've been writing, or what you've historically written.
You're going to get worse results. If you want to be a pro user, in my opinion, you reset that context window basically every single conversation just to get the best quality information. Once you get like 10 chats down into a thread on let's say a document or whatever, take your document, start over, and you'll get even better insight again.
It's still a hack that I highly recommend. Something I've been talking about in our Daily Wealth as well, which you can get for free. Just download the Meet Kevin app in the Apple or Android app store. Okay. So, HSBC thinks we're only about halfway into this sort of Sandisk supercycle. Let's go or even just the memory supercycle.
Let's go look at a little bit of analysis for some of these companies. So, one thing, and it relates to OpenAI as well, is that I published on February 1st to stock members, course members, in the Alpha report, which remember we've got a coupon expiring the end of the day. Go to meetkevin.com to check that out.
But there were memory concerns regarding OpenAI. That OpenAI has 40% of memory supply under lock. They're not just ordering finished RAM. They're also securing silicon wafer manufacturing. So, basically they want six or 900,000 wafers per month. The problem is OpenAI's commitments to capital expenditures have recently collapsed.
We've recently seen OpenAI reduce their $1.4 trillion capex spending target by 2030 down more than half to just $600 billion. Which $600 billion is still insane, still really circular, and still leading a lot of people to say, "Bro, this is absolutely nutty in terms of spend and what the expectations are, but it's a lot of money.
It's still a fraction of $1.4 trillion. And some of that might be getting chopped up because of lower expenses from OpenAI into making maybe their own memory production. That might be why you're seeing memory prices come under some pressure. But look at some of these PEG ratios at current prices. Micron is sitting as a, you know, memory manufacturer at about a 3.7 PEG, which in fairness they've got pretty good margins, but even at, you know, 2.6 if you wanted to give them a high service margin.
Even at about a 2.6, this company is sitting at about 30% overvalued right now. A Sandisk, remember when we first did our analysis on this, it was worth a whole lot less than this. It's I think it was was like a third almost of of what it is now, which is why we always like doing fundamental analysis.
Just helps you indicate when you're getting good long-term buys or long-term exits. And this is what we do fundamentally every single day in the course member livestreams. But anyway, if I take a 6.24 PEG, and again I'm generous at a 2.69, might be fair for memory in the long term here. We are looking at a company that's presently about 57% overvalued.
So, these stock prices are relatively unsustainable. Western Digital has exploded, Micron has exploded, although it's recently come down a little bit. We can see the relative strength almost on the daily chart hit oversold here. Actually, it hit Yeah, it hit 31.6, so almost oversold. A Sandisk as well.
So, we can see these have really been riding an insane ramp since the end of 2025, and they've been a great investment for the last 6 months. But we're really getting to this point where we're probably topping out. Now, why would I say we're probably topping out? Well, mostly because of memory prices, what we're seeing with memory prices.
But also, let's go look at some of their financials, and and I'm going to show you the good, the bad. So, first, if I look at a company like Sandisk, I just want to first of all applaud them. I think they're being very smart. They're actually not spending that much money on capital expenditures yet.
This is the last 6 months ending in January. They're being very conservative with how much they're expanding. Usually in a boom and bust cycle, these companies go crazy and expand their manufacturing capabilities too fast. But these companies have been through these cycles many times before, and they know if they overexpand, they're going to like rapidly crash the memory chip prices, and they'll sort of induce their own bear market, which you don't want.
So, instead, these companies have learned from the 1990s computer boom, the cloud computing boom of 2017 to 2019, the work from home boom of 2020 to 2023, that we're not going to throw a bunch of money in capex. Instead, what we're going to do is we're going to be smart, we're going to pay off our debt, and we're going to limit how much stock compensation we pay out.
All these things are happening, which is really good, really positive for this company. They have plenty of cash to pay their bills and receivables and all of Well, they have plenty of cash and receivables to pay all of their bills and all of their long-term debt. So, to me, this is a company basically with an excellent balance sheet, excellent cash flow statement.
They are literally just printing money right now. Where the risk factor comes in for this company is here, which right now looks really good. Revenue is up 61%. Their costs of goods sold collapsed to 49%, which means this company generate you know, operates at 51% margins. This is a better margin than AMD.
You are starting to knock on the door of Nvidia margins at this company. I mean, you could almost argue that Sandisk was sort of like an Nvidia 2.0 over the last 6 months, which is really impressive compared to what Nvidia did between, you know, 2020 3 and 2025. Nvidia's obviously, long topped out, and you know, I I made multiple seven figures on an investment in Nvidia when nobody wanted to buy it.
I like buying stuff when nobody's buying it. I don't like buying at pops. So, in 2022 when I was buying it, everybody was mad at me. They're like, "Oh, it's down again. It's down again." And then, you know, it was a seven-figure payday. There are plenty of stocks you could do that in now that I think we'll have those seven-figure paydays in the next probably 6 to 10 years.
Hopefully sooner. But anyway, we talk about those also over at meetkevin.com. But anyway, this right here is very interesting because their their costs are plummeting, and their operating expenses, so operating the business is only up 15.8% while their revenue is up 61%. This is really good. I mean, their net income exploded.
And a lot of this is strictly because not because they're making more chips. Obviously, they're increasing output, but not a lot. It all has to do with one thing, and it all has to do with this fancy little word called price. This is why it's really actually important to pay attention to what's happening with memory chip prices because once you stop going up like this, those crazy expansions slow down, and you kind of top out on these stocks.
So, you know, it's probably time to take the exits on some of these. Obviously, not personalized advice, just my opinion that as we start seeing cooling in memory, we're likely to see in memory price growth, we're likely to see these incredible increases slow. So, really I don't have anything bad to say about these numbers.
The only thing that I get frustrated by is not only that those prices starting to slow, but also those PEG ratios. Those the what you are paying for these companies is ridiculous right now. That's that's all. So, I just, you know, be careful. But anyway, that's my take on on memory chip pipe chip prices.
Now, as far as private credit, I do want to also provide good news because, you know, it has a lot to do with software and OpenAI. Private credit is is frankly stabilizing. I've talked about this in a video before. I want to bring it up again. Leveraged loans have stabilized since really the Iran war began.
We were essentially in freefall, and we've seen some stabilizations. Now, that doesn't mean it's good. I mean, if you look at bond prices for like Salesforce, they're trading at 58 cents on the dollar. Synopsys at 98 and 9 cents, but Aptiv at 75 cents. I've got CDK Global at 71 cents. So, you got some problems here in in terms of bond valuations, but they've stabilized is the point.
Now, JP Morgan is of the mindset that this was all based on sentiment, the guardrails of redemption caps worked, and that this is not an indicator of broad weakness. This is just idiosyncratic, and it'll it'll pass basically. And the guardrails of redemptions are working. To me, honestly, this sounds a little bit like 2007 when Ben Bernanke said the subprime mortgage crisis was contained, and you know, the pain has already been priced in.
So, we never really know how deep this stuff goes, but it's worth paying attention to charts like this because they can tell us if things are getting worse or better. And so far, it actually seems like things are getting stable, which is also what we've seen in the labor market justified by the ADP report that we got this morning in the jolts information.
So, underlying, you know, and and absent all this craziness with the Iran war, I don't think we're seeing this insane economic collapse before our eyes right now. We know certain sectors are bubbly and overinflated, but and we've seen some prices cool in the stock market, certainly nothing to insane panic levels.
Some software stocks have really sold off, which I think creates some opportunity. I think broadly, OpenAI is not a stock. Sort of bottom lines here. Open AI is way overvalued, not something I would want to touch at these prices. Anthropic probably still has some squeeze in it, but not that much anymore.
It's one of the reasons why you're starting to see, you know, SpaceX confidentially filed for IPO because it's it's time to ring the bell and take the profits, right? I mean, we got into Starlink in my venture capital fund when it was about a an Anthropic valuation. And now it's expected IPO for over you know, somewhere around four to five x that, which is insane.
Now we'll see, you know, how that actually turns up like, you know, call me when the cash is in the bank so to speak and we'll see how much money actually generated after dilution and all that cuz you never you never know how they end up structuring all this stuff. We'll see once we actually get the paperwork.
But what's really remarkable is Anthropic does sell for discount to Open AI. Anthropic is really good. Google's got a fantastic sort of sidekick business inside what is already a very good business. I mean, Google's trading for a 3.5 7 trillion dollar market cap. You know, 0.5 of that is basically probably Gemini, but I'm very impressed with Google and how they integrate their you know, products and AI services into Google admin, which is huge, especially if you're business owner.
Now if I go look at a valuation for Google just to do a quick check. Uh Google Google Google Google. Okay, let me say I'm sitting at a 1.6 peg right now. So honestly, to wrap this up beyond mentioning the coupon code, Google is cheap right now. Google should probably be selling for about, let's see here, divided by 1.61.
Okay, times about 294. Google should probably be selling close to $500. Well, that that means you have serious upside in a stock like Google. Uh now, you know, Microsoft is potentially one of those beneficiaries as well, but I mean, look at this. Google was recently at $271. And so off of highs at 294 divided by 349.
Uh we we had a little correction over here. Uh yeah, we're down about 15.6% to today's price, you know, almost 20% to the low there. This uh if if there's some volatility because of what happens with Iran over the next few weeks, be some potential over here. So yeah, it's it's a very desirable stock and they make a lot of money and you're diversified, you know, even if that Gemini becomes a smaller part of the business or commoditizes against the others, you have lower risk compared to the others.
Uh another one to watch obviously is AMD. AMD actually trades very inexpensively. AMD, as long as this AI boom keeps going, probably has a double in it. It's trading very cheaply and so was Nvidia. I mean, I kind of think Nvidia price is at top, but uh on on a technical basis and sort of a momentum basis, but on a fundamental basis, Nvidia is like a $300 stock.
So um there's definitely plenty of opportunities out there. And uh I hope you go find them. So good luck. Go make lots of money. If you want to see what we're buying, go check it out over at meetkevin.com and we'll send you alerts. usic] Meet Kevin. Always great to get your take.
Comments 0