History Is Being Made. You Only Get One Chance!

In October 2011, a 50-year flooding event in Thailand took roughly 80% of the global hard drive capacity offline. Seagate's gross margin moved from around 20% to roughly 32% in two quarters, and net profit nearly quadrupled in just a single quarter. But here's the thing, Hormuz is not a regional disruption like Thailand was. Because in addition to oil, the strait handles roughly a quarter of the global liquefied natural gas, the primary shipping route for helium out of Qatar, and a significant share of the global sulfur supply that's responsible for sulfuric acid that's used to etch circuit boards and chip wafers. And helium also cools the lasers that run every fiber optic cable inside a modern data center, and it seals every high-capacity hard drive. So when those inputs get harder to source, a meaningful part of the global technology supply chain completely stops working. The risk to producers without pricing power is severe, and the players that stand to win this setup are the ones that are positioned exactly like Seagate was in 2011. So in this video, I'm going to be walking through 10 technology companies that pass a very clean filter. Two quarter margin expansion already on the record, signed contracts with named hyperscaler customers, and production already booked through 2027. And the first six are going to be storage, optical, and semiconductor names. And the last four are just going to be AI power names, of which they all happen to be the ones that I'm watching. I want to address something right away, because a lot of these names have run very hot, and several of them are sitting at their near all-time highs. So as I walk through these, please watch for pullbacks. You probably don't want to chase these straight up, but keep this in the back of your head too. The supply constraints tied directly to these companies are very real, and they can easily push these stocks well past their all-time highs from here. Both of these things are true at the same time. So high conviction in the business, but you need to have discipline in the entry. But before I dive in, if you get any value from my videos, please consider pressing the like button so my channel can continue to grow. And of course, as a reminder, I'm not a financial advisor, and this is only for educational purposes. However, if you'd like to see my watch list, or my trades that are happening in real time, and my portfolio, I'd love for you to join the community on Patreon and Discord. Alright, the first name on the list is Lumentum, and what they do sits at the absolute physical bottom of every modern AI data center. Lumentum makes the lasers and the optical components that move data between GPUs at AI training speeds. Specifically, they manufacture externally modulated lasers, or EMLs, and ultra high-power pump lasers that integrate into the optical transceivers running between racks of NVIDIA and AMD chips. When a hyperscaler builds a training cluster, the connectivity layer tying thousands of GPUs into one logical machine is photonics, and Lumentum is the supplier the rest of the industry depends on for the highest bandwidth tier. The reason Lumentum specifically benefits from this Hormuz-style supply tightness is their structural position. The 200-gigabit per-lane EML is the component required for the next generation of 1.6-terabit optics, and Lumentum is the dominant volume supplier of that part. Per CPO Market Research, Lumentum is currently the only company shipping 200-gigabit per-lane EMLs at volume, with Coherent operating in adjacent photonic segments, and industry analysts project a 36-percent EML supply shortfall just in 2026. So when the macro environment tightens further, customers stop negotiating and start locking in capacity at the premium prices, which is exactly what NVIDIA did with a $2 billion equity investment and a multi-year purchase commitment in March of 2026. Lumentum's gap gross margin tripled from 16.6% in mid-2024 to 44.2% in the most recent quarter, and the simple reason is segment mix. The AI optics business now represents 86% of their revenue, and the legacy commercial laser segment that makes up the remaining 14% of the business is actually shrinking or going down 15% year-over-year. Next on the list is Sienna, which sells the optical transport boxes that move data between data centers when a single AI training cluster spans more than just one building. The product driving the current cycle is WaveLogic 6 Extreme, a Coherent optical platform that lets hyperscalers run a single logical GPU fabric across multiple physical campuses. 90-plus global customers have committed to it in the past two quarters, and the four largest hyperscalers have lined up over $600 billion in 2026 capex that flows directly into the boxes that Sienna makes. Sienna's specific benefit comes from the structure of the Western optical market. After Nokia had absorbed Infineera in early 2025, high-end optical networking just became a two-vendor field, with Sienna holding roughly half of the U.S. market. In a two-vendor market, when supplier costs rise across the bill of materials, those costs do not get negotiated away. Their CFO laid out the plan in the 2026 Q1 call. The company raised prices in the quarter alongside engineering cost reductions, and they guided to additional price increases kicking in during the second half of the year. The Q2 results just came out, and Sienna grew their revenue 40% year-over-year, accelerating from 33% the quarter before, and adjusted earnings per share nearly quadrupled. Management raised their full-year guidance to $6.3 billion, which works out to be about 32% growth for the year. And even after all of that, the market sold the stock down hard because expectations going in were just ridiculously high. And that pullback dropped right into my buy zone. You've probably noticed the quant ratings and the earnings transcripts that I've been pulling up for the first couple of names. That all comes from one place, and that brings us to today's sponsor, Seeking Alpha Premium. Now, I have built my own research process over the years, but Seeking Alpha earned a permanent slot in my workflow, and I've been using it now for the past two years. And there are two features that I lean on the most. The earnings call transcripts that let me search exactly what management said about guidance without having to sit through a full 90-minute call. Take Palantir, for example. On their latest call, I pulled the line where the CFO raised their full-year revenue guidance to the $7.65 billion range, which happened to be their largest raise ever in all of about 10 seconds. And the quant ratings I happen to use as a data check against my own read, especially where the model disagrees with me because it's sometimes more important than just when it agrees. And the methodology holds up because Seeking Alpha's back-tested quant strategy beat the S&P 500 every single year for the last 12 years. Use the link in the description to get $30 off your first year at a 7-day free trial. If you're looking for serious help with your research, this is a great place to start. Third on the list is Seagate Technology. And this is the name where the Thailand parallel is obviously going to be the most direct. Seagate makes the high-capacity hard drives that hyperscalers use for near-line storage. The data they want fast, but not GPU fast. Their Mosaic platform is the only heat-assisted magnetic recording drive shipping at scale, with two hyperscaler customers already qualified. So roughly 87% of their shipments go to cloud and data center customers. And here's the thing, every drive above 10 terabytes is helium sealed. That helium dependency is what ties Seagate directly to the Hormuz situation. Qatar's complex produces roughly 30-38% of the world's helium supply. And Qatar Energy declared force majeure on the facility in March 2026 after the natural gas halt cascaded down into the helium extraction. This outcome resulted in Seagate having a very direct response. The company raised production allocation pricing by 20-30% and finalized bill-to-order contracts with hyperscalers through fiscal 2027. So what the financials show is the largest two-quarter margin move across all of these 10 names. Gap gross margin expanded from 41% in December 2025 to 46.5% in March of this year. And their future guidance for 2026 puts gross margin above 50% for the very first time. This is the 2011 Thailand pattern playing right out in front of us, only it's a little bit cleaner this time. Because in 2011, Western Digital lost share when its Thailand fabs completely flooded out. But in 2026, both of these companies are going to benefit from that same helium-led recovery. Which brings me to the fourth company on the list, which is Western Digital, which happens to have the cleaner balance sheet version of the same Seagate trade. Western completed a full spin-off of its SanDisk Flash business in February 2025, leaving the parent company as a pure-play hard drive maker for the very first time in its modern history. Roughly 89% of revenue now comes from cloud and hyperscaler customers. And the technology path differs from Seagate. Where Seagate is leaning on heat-assisted magnetic recording, Western Digital is leaning on ultra-SMR, which is a higher-density approach that's already shipping 40TB volume drives in early 2026. Western Digital captures the helium constraint through this same channel as Seagate, with the same hyperscaler customer base on the receiving end. But what Western Digital did with that setup is lock it down a little bit longer. 5 of the company's top 7 hyperscale customers are now on a multi-year long-term agreement through 2026. And one of those committed through calendar 2027. And select agreements extended into 2028 and also 2029. They raised average selling prices roughly 9% year-over-year on the most recent quarter alongside the build-to-order book. Western Digital also crossed 50% gross margin for the very first time in its history in Q3 of this year. And the segment data shows the milestone is essentially a cloud business achievement because the cloud segment is now 89% of revenue and grew 48% year-over-year. Combined with the cleanest balance sheet across these 10 names, that is what makes Western Digital the cleaner side of the same Seagate trade. 5th on the list is Credo Technology Group, and it's the only name in this video whose business gets stronger when the helium and laser supply chain breaks even further. Credo makes the high-speed copper cables that connect AI chips and switches inside a data center rack. These are called active electrical cables. For long-distance connections, fiber optic cables win hands-down on power efficiency, which is why fiber dominates the internet backbone. But for short connections inside a single server rack, copper actually wins, because every fiber link needs a laser and the power-hungry electronics that convert electrical signal to light and then back again. Credo's smart copper cables skip that conversion entirely, using roughly half of the power of a fiber transceiver at short reach. And Credo holds about 73% of that market. Now, here's why Hormuz drives demand. Every short-reach fiber link in a data center needs a laser, rare materials, and helium to keep the laser cool. When those inputs get harder to source because of Hormuz's constraint, hyperscalers can't really get enough fiber transceivers at any price. They are essentially forced to architect more of their builds around Credo's copper for every link where copper works. Now, this isn't exactly a temporary substitution. It is a redesign of how the racks get wired. And Credo is the structural beneficiary on both the cost side and the demand side of that shift. In the most recent fiscal quarter, Credo tripled its revenue compared to the same quarter a year earlier, and gross margin expanded to over 68%. If you were to look back a year ago, the company was losing money. But in their most recent quarter, they were earning almost 39 cents of net profit on every dollar of revenue. That's the exact type of trajectory of a company on the right side of the trade. Our next company is Astera Labs. And Astera is the chip company that benefits when hyperscalers redesign their racks away from fiber. Astera makes three kinds of connectivity chips that sit between AI processors. The first is a retimer, which keeps high-speed signals clean as they travel through copper inside of a rack. The second is a smart cable product that competes head-to-head with Credo. And the third is a fabric switch that connects many GPUs into a single logical machine. NVIDIA, AMD, Microsoft, Google, and Meta all use Astera Silicon. And of course, Astera holds a board seat on Open UALink Standard with the hyperscalers built specifically so they can avoid being trapped by NVIDIA's standards. Now, Astera's angle on all of this is rack redesign. When fiber-optic connectivity gets really expensive and constrained, hyperscalers do not want to reduce the numbers of GPUs that they're ordering. So instead, they redesign the racks, packing more processors into a single rack or a pod, and connecting them with copper and short-reach links. Because that side of the supply chain is not affected by Hormuz. And every time a hyperscaler does that, the chip content per rack goes up. Meaning more retimers, more fabric switches, and more smart cables. Oh, and guess what? Astera makes all three. And their new 320-lane fabric switch, which just started shipping in volume in May of this year, is specifically designed to support exactly that redesign. Most recently, Astera grew revenue 93% year-over-year, while expanding gross margin That is a fabulous chip designer growing at hyperscaler speed, with gross margin above three-quarters of revenue. A combination that does not exist anywhere else in this video. Now we're going to move on to the AI power side of the list, starting with Constellation Energy, which operates the largest nuclear fleet in the United States. 21 reactors with 22 gigawatts of At the beginning of the year, the company closed a $16 billion acquisition of Calpine, adding 33 gigawatts of natural gas and geothermal generation. Combined, Constellation is now the largest private-sector power producer in the country. Its biggest customers are the hyperscalers building AI data centers everywhere. Microsoft signed a 20-year contract for the restored Three Mile Island. Meta signed a separate 20-year contract for the entire 1.1 gigawatt output of the Clinton, Illinois plant. Both, of course, are at hyperscaler premium pricing, and their connection to Hormuz comes through natural gas. The Strait carries roughly a quarter of the world's seaborne LNG, and when global natural gas supply tightens, U.S. natural gas gets pulled overseas to fill that gap. And of course, U.S. gas prices rise as a result, which pushes up electricity prices on the grid because gas plants set the going rate for most of the day. Constellation's nuclear plants run at a fixed marginal cost of around $32 to $35 per megawatt hour, which makes them the cheapest source of power on the grid. And of course, when the price rises because of gas, the nuclear fleet captures the entire spread, more revenue and more margin. The big thing that changes for Constellation is their customer base. Five years ago, the company was a regulated utility with slow-growing investments. Today, it's the dominant counterparty for the hyperscalers that are chasing 24-7 carbon-free power. With two 20-year contracts already locked in and a combined fleet that has more than doubled, their management team is already guiding 2026 earnings up nearly 20% year-over-year simply because of the strength of Calpine integration and the rising power prices. And right on the heels of Constellation comes Talon Energy, the concentrated, single-asset version of the same trade. Talon operates the Susquehanna nuclear plant in Pennsylvania with 2.5 gigawatts of generating capacity. In June of last year, the company signed a 17-year agreement for $18 billion with Amazon Web Services. That single contract covers 76% of Susquehanna's entire output. Talon also closed two natural gas plant acquisitions in They signed a third $3.4 billion gas plant deal in January of this year, building out a multi-gigawatt dispatchable fleet alongside the nuclear asset. Talon captures the same gas price spillover that benefits Constellation, but through a different lever. The Amazon contract is fixed price for the first 10 years, and then resets to a fixed margin above market prices for the remaining seven. So with natural gas, prices stay elevated because of Hormuz pressure through the late 2020s and into the 2030s. Talon does not have to renegotiate to capture that upside, because the reset is going to be automatic, and the margin rides the energy and capacity prices wherever they go. The real driver is the operational activation that's happening in spring of this year. Talon is in the middle of a refueling outage in Susquehanna, and the transmission reconfiguration that turns Amazon contract from announced to in-service is being completed at the same time. In their latest quarter, their revenue had grown 79% year-over-year, and adjusted free cash flow quadrupled. The Amazon contract has not yet started delivering its full economics, and the full ramp continues through 2032. Now the third name on the power side of the list is Bloom Energy, and Bloom is the one that bypasses the grid entirely. Bloom makes solid oxide fuel cells. These are not the backup generators that sit idle in a hospital basement. They happen to be the small power plants that run 24 hours a day, converting natural gas or hydrogen directly into electricity through an electromechanical reaction at the customer's location. The customer simply plugs the data center into the fuel cell as a primary source of power, and they don't even touch the grid. Bloom is the only fuel cell maker shipping at gigawatt scale, and they have signed three transformative contracts in the past year. Oracle even locked in a master agreement for up to 2.8 gigawatts. And AEP signed a $2.6 billion agreement covering Amazon and another customer. And Brookfield committed $5 billion to finance further Bloom deployments at AI sites. So where Constellation and Talon capture Hormuz through gas price spillover, Bloom captures it on the demand side. Connecting a new AI data center to the grid takes five to seven years because every US utility is already overwhelmed by AI capacity requests. So when you layer on Hormuz volatility on top of that bottleneck, hyperscalers really stop waiting and they start acting. The obvious choice is just to build their own power on site. Bloom can ship a working power plant to a data center within a few months, not having to wait years. The Oracle, AEP, and Brookfield contracts are just proof that the largest AI buyers in the world have made their decision and started to commit at scale. After a decade of losses, Bloom flipped to positive operating income in 2024, and they've been accelerating ever since. In their latest quarter, their revenue has grown 130% year-over-year, but their product line grew 208% on its own, and their product line runs at a 35% gross margin versus 18% for services. So that mixed shift to product is driving both the top-line growth and the margin expansion. Now we'll move on to the next name on the list, which is Powell Industries, and it happens to be the small-cap pick within the list. Powell designs the custom-engineered electrical equipment that sits between a power source and a builder's actual circuits. Their core products are switchgear, the high-voltage devices that route and protect electrical flow, and bus duct and heavy-duty conductor that carries large amounts of electricity inside of a facility. When a hyperscaler builds an AI campus, somebody has to engineer the substation equipment that connects to the data hall, and Powell is one of the only specialists that they can go to for that type of work. On the data center side, the company secured a $400 million mega order for AI data center switchgear. That single order is the largest project in Powell's history. And on the oil and gas side, when the Strait of Hormuz creates risk to international natural gas capacity, the structural case for U.S.-based LNG exporters begins to strengthen. And those exporters need exactly the kind of switchgear that Powell sells. Oil and gas already represent 29% of Powell's backlog, and management has explicitly tied the outlook to U.S. exporter advantages from the international supply constraints. The segment data shows where Powell's growth is really coming from. Electric utility revenue, the bucket where utility route and data center work gets billed, grew 49% year-over-year. And in oil and gas, the legacy industrial segment, it stayed relatively flat. And the results are really showing up in their backlog, where it had started at $1.8 billion, and it's already been bumped up to $2.2 billion. To boil it all down, the AI grid work is completely taking over their business. That's a good thing. Now, that happens to be the 10 names. The Hormuz constraint flows into the technology supply chain through three different channels. The helium and rare materials that go into lasers and high capacity hard drives travel straight through the Strait. And that flow is what drives the movers at Plumentum, Sienna, Seagate, and Western Digital. That same constraint creates demand for the copper and chip level alternatives that bypass the laser supply chain entirely, which is what Credo and Astera Labs are doing right now. And the natural gas that sets electricity prices gets pulled overseas when global LNG begins to tighten, which is the mechanism driving margin expansion at Constellation, Talon, and Bloom Energy on the power side. Powell happens to sit at the intersection of both, selling the switchgear that goes into AI data centers and US LNG export facilities at the same time. So every one of these companies is showing two-quarter margin expansion with demand locked in for the next year. As I said at the beginning, that's my filter. The companies that pass it are the ones positioned to convert this constraint into margin, not just absorb it. As always, thanks so much for watching, and I'll see you next time.