# The AI Trade Is Changing... These 3 Stocks Could Explode Next

https://www.youtube.com/watch?v=BmpdgX7Ylpg

[00:00] Last month, Anthropic made a big move
[00:02] that you may have missed. Its cla is now
[00:05] living inside of Excel, PowerPoint, and
[00:08] Word. It is generally available. This is
[00:10] not a demo. There is not a wait list. It
[00:13] is now embedded in three apps that 1.4
[00:17] billion office workers open every single
[00:19] morning. Microsoft there, too. Their
[00:21] co-pilot is around 20 million paying
[00:23] users, and they rolled out a new Office
[00:25] bundle on May 1st. And here's why that's
[00:29] a bigger deal than it sounds. Up until a
[00:31] few months ago, if you said you were
[00:33] using AI at work, that meant you're a
[00:35] developer. You're writing code in
[00:37] something like Cursor. That's maybe 20
[00:40] 25 million people on the whole planet.
[00:42] But now AI is about to operate the
[00:45] actual software that over a billion
[00:47] knowledge workers use to do their jobs.
[00:50] This is a major inflection point that
[00:52] has real ramifications to the supply
[00:55] chain underneath it. Today I'm going to
[00:56] show you exactly what Agentic AI is
[01:00] about to do to global computing demand.
[01:02] Why it is a completely different animal
[01:05] than the chat bots you've been using.
[01:07] And then I'm going to hand you a basket
[01:09] of backdoor stocks that quietly power
[01:12] this entire buildout. Some of these
[01:14] you've never heard of, and that's the
[01:16] whole point. Make sure to subscribe to
[01:18] the channel cuz we are at the very
[01:19] beginning of this trend. And I'm going
[01:21] to keep you steps ahead of the average
[01:23] investor the whole way through. Now, let
[01:26] me break this down first with an
[01:27] admission. Back in 2023, I wrote a piece
[01:30] basically warning people that Nvidia
[01:32] looked expensive, that it was
[01:34] overpriced. It had tripled in a few
[01:37] months. The price to sales ratio had
[01:39] doubled. I looked at that valuation. I
[01:41] thought it was silly. And well, I was
[01:45] wrong. Because while I was staring at
[01:47] multiples like some Warren Buffett
[01:49] wannabe, the actual sales were about to
[01:52] explode. Nvidia's revenue went from 61
[01:55] billion to 130 billion to 216 billion in
[02:00] just three years. And doubling every
[02:03] year at that scale is frankly something
[02:07] I had never seen before. But every big
[02:10] tech company on Earth was in an arms
[02:12] race to buy GPUs and Nvidia was the only
[02:15] one selling. So what did I learn from
[02:17] that? Don't bet against a structural
[02:20] change. Now, if you haven't started
[02:22] playing with all the fancy AI tools yet,
[02:25] you need to understand this. An AI agent
[02:28] is not a chatbot. All right? When you
[02:31] ask chat GPT a question, it's quick. It
[02:34] doesn't use a lot of compute power. But
[02:37] AI agents are totally different beasts.
[02:40] So, compute is measured in tokens. So, a
[02:43] developer running a focused 30 minute
[02:45] coding session burns somewhere between
[02:48] 50 and 200,000 tokens. Now compare that
[02:51] with what I might do as a financial
[02:53] analyst with agentic AI. I might ask an
[02:56] agent to analyze the nuclear power
[02:58] ecosystem. Analyze every publicly traded
[03:01] company in the space. Build and
[03:03] reconcile a 200 row file. Compare it to
[03:06] industry comps. Write a full email
[03:08] summary. And then build me a 40 slide
[03:10] deck from that information all from one
[03:12] command. So, as opposed to 50 to 200,000
[03:16] tokens every half hour for a programmer,
[03:19] that one task would burn 800,000 to 2
[03:22] million tokens. Now, multiply that by
[03:27] 1.4 billion office workers.
[03:30] Microsoft has already said over 80% of
[03:33] the Fortune 500 has rolled C-pilot out
[03:36] to part of their workforce. The load
[03:38] this puts on the system makes everything
[03:40] that came before it look like a rounding
[03:43] ear. And the infrastructure to handle it
[03:46] does not exist. Not yet. Remember, a
[03:49] chatbot session is short, GPUheavy, low
[03:53] context. An agentic session is the
[03:55] opposite. It could run for hours. It
[03:58] constantly does things and it leans on
[04:00] CPUs, memory, and GPUs all at once. So
[04:05] the big AI boom calls everyone to go to
[04:08] GPUs. That's what Nvidia makes. And the
[04:10] the server CPU to GPU ratio, stay with
[04:15] me for a second, has shifted from 1 to
[04:19] 12, 12 being the GPUs at the peak of AI
[04:22] boom in the old training builds to 1 to2
[04:27] in these new agentic builds. So you've
[04:30] got this whole pile of components.
[04:32] Nobody was talking about CPUs, memory,
[04:35] storage, power chips, stuff they didn't
[04:37] care about anymore. That all became
[04:38] missionritical all at once. The picks
[04:41] and shovels layer just got a massive
[04:43] promotion. That's why stocks like
[04:45] SanDisk, like Micron are going vertical
[04:48] right now. Get this. Goldman Sachs now
[04:51] models that consumer and enterprise
[04:53] agents will push monthly token
[04:55] consumption, again that's how much
[04:57] compute, to roughly 24 times the entire
[05:01] world's current capacity by 2030. In
[05:05] that same report, Goldman noted that
[05:07] quote tooken economics turned positive
[05:10] in the first half of 2026. That is the
[05:13] moment an agent's output is worth more
[05:16] than what it costs to run it. And once
[05:19] that flips,
[05:21] every company on Earth has to deploy
[05:23] agents just to stay competitive. Right
[05:25] now, Agentic AI is still more hobbyist
[05:27] tinkering than full commercial
[05:29] deployment. But the second it pays for
[05:31] itself, adoption goes vertical, months,
[05:34] not years. And get this, we are still
[05:36] ridiculously early. The latest estimates
[05:39] say about 78% of humanity has never used
[05:43] a single AI tool. Not once. Now, by the
[05:45] way, if you like content like this and
[05:47] you want the actual trades I'm taking,
[05:49] you need to join my Black Ops trading
[05:51] service. For just five bucks, you get an
[05:54] entire year of access. Every Monday, I
[05:56] go live for an hour and walk you through
[05:58] my real positions, the stocks I'm
[06:00] watching, buy and sell rules. This is
[06:02] real money, uh, real trades live there
[06:05] with you every week for a year. You'll
[06:07] also get my weekly newsletter,
[06:09] proprietary Trading View indicators,
[06:11] bonus reports, direct access to the
[06:13] team, ton of stuff, just five bucks. So,
[06:15] click the link in the description, scan
[06:17] the QR code, or go to tradewith ross.com
[06:20] to get signed up. Now, how do you
[06:23] actually play this? How do we take
[06:24] advantage? Now, a lot of the best names
[06:27] in this stack, the best stocks and
[06:30] supply chain trade overseas. And I'll
[06:32] get to those cuz some of them are real
[06:34] hidden gems. But I know most of you want
[06:36] something you can buy this afternoon in
[06:38] the accounts you already have without
[06:40] jumping through a bunch of hoops. So let
[06:41] me start with three USlisted names that
[06:44] each own a different layer of the
[06:47] agentic build. CPUs, memory, and
[06:50] storage. All right, a heads up before I
[06:52] do. These three have already run. Some
[06:54] of them hard. These are not sleepers
[06:56] that nobody noticed. These are the
[06:58] proven leaders of this wave. But they
[07:00] are the cleanest way for a US investor
[07:03] to own the picks and shovels underneath
[07:05] Agentic AI and the demand wave driving
[07:08] them is still early. Plus, we have
[07:11] started seeing some pullbacks thanks to
[07:12] a few dips in the market here recently.
[07:14] Stock number one, ARM Holdings, ticker
[07:17] ARM. Now, this is essentially the toll
[07:20] booth. When one of these big
[07:21] hyperscalers builds its own custom AI
[07:23] chip, so Amazon's Graviton, Google's
[07:26] Axion, Microsoft, Nvidia, etc., Almost
[07:28] every one of them is built on ARM's
[07:31] architecture. And ARM collects a royalty
[07:33] on each and every chip. They don't care
[07:36] who wins the design more. They don't
[07:38] care if it's Amazon or Google or
[07:39] Microsoft. They're not making a bet.
[07:41] They get paid either way. And that is a
[07:43] business I want exposure to. A royalty
[07:45] stream on the entire custom CPU boom
[07:49] listed right here on the NASDAQ. Ticker
[07:51] ARM. Stock number two, Micron, ticker
[07:54] MU. Now remember what I told you about
[07:56] Agentic AI agents run for hours and they
[07:59] are hungry for memory. That's where all
[08:01] the context lives. Micron is the only
[08:04] US-listed pure play in high bandwidth
[08:07] memory. That's the stuff that sits right
[08:09] next to the GPU. And here's the tell.
[08:12] Their entire 2026 HBM production is
[08:15] already sold out the whole year. The
[08:18] stock has gone parabolic. It just
[08:20] crossed a trillion dollar valuation. So
[08:22] respect again. It is extended, but so is
[08:25] Nvidia a couple of years ago, and that
[08:27] one kept going as well. And when a
[08:29] company sells out an entire year of
[08:31] capacity in advance, you know, the
[08:33] demand is big, and it is real. Stock
[08:35] number three, Seagate, ticker STX. Now,
[08:37] agents generate mountains more data than
[08:40] chat bots ever did. And all that data
[08:42] has to live somewhere. Seagate is one
[08:45] half of the hard drive duopoly. And for
[08:48] the first time since 2017, they've got
[08:50] real pricing power. capacity reportedly
[08:52] sold out through year end. So if you
[08:55] want the storage layer of this trade,
[08:57] this is the easy US way to play it. And
[09:00] if you want to go deeper into storage,
[09:02] it's Duopoly partner Western Digital,
[09:04] ticker WDC and ND, which is the spin-off
[09:08] of SanDisk. There's both right there on
[09:10] the NASDAQ, too. Okay. Now, here are the
[09:12] names you haven't heard of, the ones
[09:13] Wall Street's not talking about because
[09:15] most of them don't trade in the US. But
[09:17] just because it's not a US stock doesn't
[09:19] mean it's not a great opportunity. On
[09:21] the Tokyo exchange, you've got Renis,
[09:24] ticker 6723. Yes, their tickers are
[09:26] numbers. I know is what it is. They're
[09:29] the power management chips. Nidobasiki
[09:32] 3110. That's that 70% glass cloth
[09:36] monopoly. I like you've got Marada 6981.
[09:39] They're the number one capacitor maker.
[09:41] Kioxia 285A, fresh IPO in the NAND
[09:45] space. In MEC, ticker 491. They do micro
[09:49] etching chemicals. Look, I know this
[09:51] stuff is like, what is he talking about?
[09:52] I'm telling you, these are the stocks in
[09:54] this layer. There's also a few in the
[09:56] Taiwan exchange. I won't even try to
[09:57] produce them or pronounce them, but
[09:58] they're up here on the screen. So, how
[10:00] would you actually play these if you
[10:02] wanted to invest in these Asian
[10:04] companies? Well, if you're trading in a
[10:06] regular individual or joint account,
[10:08] it's pretty simple. Open one up at
[10:10] Interactive Broker. IBKR gives you
[10:12] direct access to the Tokyo and the
[10:15] Taiwan exchanges. You can buy these in
[10:17] US dollars alongside your US stocks. No
[10:20] issue. Schwab's global account works
[10:23] too, but heads up, it covers Japan only,
[10:25] not Taiwan. So, it's not going to get
[10:26] you all these. Uh, Fidelity also lets
[10:29] you trade in Japan, but only in
[10:30] non-retirement accounts. Now, if you
[10:33] want these in an IRA or a Roth, that's
[10:35] going to get pretty tricky. Most brokers
[10:37] aren't going to let you do that. There
[10:38] is a workaround, however. A handful of
[10:41] these have what is called an ADR. stands
[10:45] for American depository receipt. It is
[10:48] basically it's an IOU for a share of
[10:51] that stock on the foreign exchange but
[10:53] it's a standin that trades on the US
[10:56] exchange during our hours. Okay. So
[10:58] Renaissance for example trades over the
[11:00] counter is RNY.
[11:03] Marada is M R A Y. Foxcon is HNHPF.
[11:08] These three you could buy in an IRA in a
[11:11] regular brokerage account. anything with
[11:13] US dollars without access to
[11:15] international markets. But for the rest
[11:17] where there's no good ADR, at the end of
[11:19] the day, there's not a whole lot you can
[11:20] do. I would stick with the US listed
[11:22] leaders, the ones with ADRs here, the
[11:25] ones we've already covered, especially
[11:26] ARM, Micron, Seagate. That's going to
[11:29] give you clean exposure to the same
[11:31] trend without ever leaving NASDAQ listed
[11:33] stock. Now, there are obviously risks.
[11:36] First, more than half this basket sits
[11:38] in Taiwan and Korea because that's where
[11:40] the supply chain lives, where they're
[11:41] making the stuff. So some big
[11:43] geopolitical event in Taiwan would hurt
[11:45] the basket of stocks. Also, six
[11:47] customers, all of which are names, you
[11:49] know, Google, Amazon, the usuals,
[11:51] account for over 70% of the orders
[11:53] across most of these names. So if two of
[11:55] them pause spending, sales take a hit.
[11:57] It is doubtful, but possible. Valuations
[12:01] are also rich. But as we learned with
[12:03] Nvidia a few years ago, often times
[12:05] those rich valuations are justified. The
[12:09] demand here is structural. The moes are
[12:11] the best in the business and the
[12:13] buildout underneath Agentic AI is only
[12:15] just beginning. I believe pullbacks in
[12:18] names like these are generally going to
[12:19] be buying opportunities. Look, we know I
[12:23] keep talking about it. The Nvidia trade
[12:25] is one everyone already made. The
[12:27] company's worth 5 trillion bucks now.
[12:28] It's up 1 1600%. Folks, it happened.
[12:32] It's over. But the agentic AI trade, the
[12:35] memory, the storage, the substrates, the
[12:37] sockets and power chips. This is the one
[12:40] Wall Street and general investors are
[12:41] just waking up to. Folks, don't forget
[12:44] to subscribe to the channel. And again,
[12:45] if you want to trade alongside me, come
[12:47] join my next live trading session this
[12:50] Monday. Become a member of the Black Ops
[12:52] service. Links in the description, just
[12:54] five bucks. I'll walk you through
[12:55] exactly where I'm putting my money.
[12:57] Until then, I'll see you in the next