Dips Don't Last – 8 Stocks I’m Buying

[01:00:00:04 - 01:05:00:23] Back in June, in just a matter of days,   the chip index fell about 10%. The nuclear stocks  that had tripled gave back 30%, and the quantum   names got cut nearly in half. And it looked like  the AI trade was finally breaking. Then Micron had   its quarterly earnings, where it reported revenue  up 346% from a year ago, its best gross margins   ever near 85%, and its most advanced AI memory  already sold out through 2026. With the shortage   running all the way into 2028. Those are clear  signs that the AI build out is actually speeding   up, not a bubble beginning to cool down. So these  are the top stocks that I'm watching right now.   The best companies in that whole build out that  got dragged down with everything else and now have   the most room to grow. And at the very end, I'm  going to show you exactly how I'm going to size   each one. But before we jump in, if you're getting  any value from my videos or my research, please   consider pressing the like button and I'd love  it if you'd consider subscribing. And if you're   interested in my trading activity, my portfolio,  and any of my deeper analysis, then hey, I'd love   for you to join our community on Patreon. Instead  of starting with chips, like I normally would,   instead I'm going to start with electricity,  where it all begins. And the first name is Eaton,   the power management company that builds almost  everything an electron passes through on its   way from the grid to the server. The switchgear,  the transformers, and the backup power that keeps   data centers alive. And here's the number that  completely reframes this whole company. Eaton's   electrical content runs about $3.4 million for  every megawatt of AI compute that gets built.   And the industry is on track to build roughly 13.6  gigawatts of new data centers this year alone. So   that's about $46 billion of electrical equipment  up for grabs from a single year of construction.   And Eaton's entire company did about $27 billion  in revenue just last year. So in just one year of   data center build out, it opens up a market bigger  than the whole business was last year. And this is   already showing up because Eaton's data center  orders grew about 240% over the past year. And   it's carrying the better part of a year of sales  in backlogs. Then there's the part that looks   alarming until you look a little bit closer. Last  quarter, the headline operating margin dropped   to about 15%, which looks like the business is  beginning to crack. But that was the cost of the   acquisitions Eaton was closing. Underneath it, the  core electrical business ran a 22.7% margin with   record adjusted earnings and raised guidance. Then  you need to go ahead and take a step back because   the transformation happens to be the real story.  Six years ago, Eaton was a 13% margin industrial,   doing about $21 billion in revenue. And today it  does over $27 billion at nearly 20%. So even as   the top line grew, the wider margin almost doubled  its profit. It truly earns a premium for that now.   And a pullback, like the one that we just got,  is the window to start building the position.   Now staying with power, the second name is BWX  Technologies. And this is the one nuclear stock   that I would actually buy on this pullback because  it is the only profitable one in the group and   the sole manufacturer of the reactors that run on  every US Navy submarine and the aircraft carrier.   So when nuclear sold off the gap showed up real  fast. Because the hyped pure plays like Oklo and   NuScale, well they came down about 30% while BWXT  gave back only roughly 17% and from a base that   had been climbing while they had fallen. And the  reason is simple. BWXT already builds reactors and   they get paid for it while the other pure plays  are still years away from their first one. BWXT   is actually building and the others are still  betting they're not quite there. But right now,   the Navy is just the foundation because BWXT  quietly runs four different nuclear businesses   all under one roof. They're building the first  reactor pressure vessel, the single largest piece   for one of the first small modular reactors  going up in North America. It also makes the   Triso fuel for Kairos Power, the company building  a reactor to help power Google. And its medical   arm produces isotopes for cancer treatment. From  Therasphere for liver tumors to Actinium 225,   one of the most big ones being made today. So it's  true, you are buying a defense monopoly, a piece   of the nuclear renaissance, an AI power supplier,  and a cancer treatment business all in one. And   their work is book solid because BWXT carries a  record $8.6 billion in backlog, that's up 77% in   a single year, against a company that does under  $4 billion in revenue. So it has more than two   years of work already locked in. And you can see  it flowing through to cash, because free cash flow   went from about $46 million in 2023 to nearly  $300 million in 2025. And it's guided higher   again this year. The only real catch is the price  because it's near 43 times forward earnings on a   low double digit grower. So the quality is not  the question, the entry is. So a pullback, like   the ones we've been seeing the past few months,  is really how we get better entry options.   [01:05:00:23 - 01:06:06:03] Now, if you're anything like me,   you're spending a lot of time trying to figure  out which names actually deserve a spot in your   portfolio for the back half of the year. And  that's where today's sponsor, Seeking Alpha,   comes into play. Where Seeking Alpha happens  to have a platform that I use to pressure test   almost every stock that I research. Their quant  rating system grades thousands of stocks on value,   growth, profitability, momentum, and earnings  revisions. And that back-tested quant strategy has   beaten the S&P 500 every year for over a decade.  I genuinely lean on them for their earnings call   transcripts, the top-rated screens, and the quant  grade before I even begin to build a position. And   here's exactly why I'm bringing this up right now.  On Tuesday, July 14th, from noon to 1.30 Eastern,   Seeking Alpha's Steven Kress is hosting a live  top stocks event for the second half of 2026.   These happen to be his quant-backed picks  built for growth and long-term performance.   This one is members only for premium, alpha picks,  and pro. And right after it wraps up, Steve's full   list of second half picks are going to drop as  a members-only report. So if you want to see it   at that event and the report that follows, then  join through the link down in the description.   [01:06:06:03 - 01:16:54:12] Now we move on to chips, starting   with the name that everyone expects, and that's  Nvidia. And the real monopoly that gets overlooked   is from software, not the chips. Everything in  AI is written on Nvidia's platform called CUDA,   with around 6 million developers building on  it, and over a billion of its chips already   out in the world. And that lock-ins is what no  competitor can really break. But what makes it a   monopoly in several fields at once is that Nvidia  runs the same playbook pretty much everywhere.   In the data center, it doesn't sell a chip. It  sells the entire rack as one computer. And its   networking arm, alone, tripled in the past year to  nearly a $60 billion dollar pace. And in robotics,   it builds the platform the humanoid robots are  trained in and run on every day. And in quantum,   it makes the software those machines run on, too,  with 17 of the top quantum hardware makers and   9 national labs already plugging into Nvidia. So  whoever wins AI or whoever wins robots or whoever   wins quantum, they all still pay Nvidia. Then  there's a number that completely stops me in my   tracks. Because over three years, revenue went  up eight times, from $27 billion to over $215   billion. And the stock actually got cheaper as it  climbed. Its price-to-earnings ratio falling from   over 100 to about 30. Because earnings grew faster  than the price. And that's also why the peg ratio,   which measures the price that you pay against how  fast earnings are growing, sits near a 0.5. And   anything around a one or lower tells you the stock  is cheap for its growth. So after this pullback   oversold on its 200-day line, that is exactly  where I want to own the company that the whole   AI economy runs on. Next up is the other side  of the trade with Broadcom. Because Nvidia is   the company that everyone pays. Strangely enough,  Broadcom is the one that they call when they want   to stop paying it. When Google, Meta, OpenAI,  or Anthropic wants to build its own AI chip to   get out from under Nvidia, it co-designs that  chip with Broadcom. And Broadcom also sells the   networking that wires those chips together. So  it gets paid twice on the very same cluster. And   that AI business is accelerating, not slowing,  because its AI revenue grew more than 140% just   last quarter. And it is on pace to nearly triple  this year. With management guiding it to nearly   double again the year after, all backed by a $73  billion order book. Which is why that one line   is now nearly half the entire company. And the  VMware deal everyone thought would weigh down the   company? It did the complete opposite. Because  VMware's software runs at a richer margin than   Broadcom's own chips. Around 77% against a 58%.  So it actually lifted the blended margin instead   So even though Broadcom looks expensive at around  60 times earnings, that is the VMware accounting   hiding the real number. Because on next year's  earnings, it trades closer to 24 times. With a   peg of about a 0.5. And that's just as cheap as  its growth as Nvidia. So after the pullbacks we've   been seeing back down its 200 day line, that is  my entry into the arms dealer of the whole AI chip   war. And that now brings us to the cheapest and  most beaten down of the Megas, Alphabet. Designing   its own chips, called TPUs, so it isn't paying the  Nvidia tax the rest of the industry already pays.   And it's building its own Gemini models on top  of everything else. The thing to understand about   Alphabet is that it's really for businesses moving  at very different speeds. The cash cow is search,   still by far the biggest piece, and far from  dying, its growth has nearly doubled over the past   year. Climbing from 10% to 19%. As AI mode crossed  a billion users and began monetizing questions,   it never could before. The new rocket is Google  Cloud, where revenue grew 63% and the operating   margin roughly doubled in about a year and a half.  From the high teens to 33%. So a unit that used   to lose money is now a serious profit engine.  And customers have already signed for around   $460 billion worth of future cloud work. A backlog  bigger than all of Alphabet's revenue in a year.   And quietly, underneath all of them, YouTube has  grown bigger than Netflix and become the most   watched service on American televisions. And Waymo  won't lay this ride along as a free call option,   already running more than 400,000 paid robo-taxi  rides a week. That's about 10 times what it was So   once again, it's four engines, and the fastest,  highest margin ones are taking a bigger slice   every single quarter. Now there is one thing to  flag, and that's that last quarter's profit looked   bigger than it truly was. It was inflated by a  one-time paper gain on an investment. So strip   that out and the real earnings are just a little  bit lower. Even then, it is the cheapest mega in   the group, around 24 times forward earnings with a  peg near 1.3. And it's the most oversold name here   in the pack after we've seen those pullbacks the  past few weeks. Granted, it's slowly building up   momentum, but it still has a long ways to go, and  this is where I like to see it. Now another of the   mega anchors is Arista Networks, the company that  builds the high-speed switches that wire all those   GPUs together inside an AI data center. And like  Broadcom and Google, its real edge is software,   because every Arista switch from a campus closet  to a giant AI spine runs the same single operating   system, which is why the cloud standardized on it.  For years, the standard way to connect GPUs was   NVIDIA's proprietary Infiniband, and now the whole  industry is moving that job onto Open Ethernet,   which is Arista's home turf. Meta proved it works  by running a cluster of more than 24,000 GPUs on   Arista Ethernet and matching Infiniband. And that  is why Arista's AI networking revenue doubled last   year and is guided to more again this year. And  a lot of that demand is already paid for because   customers prepay Arista ahead of delivery. And  that deferred revenue has roughly doubled to   more than $6 billion just sitting there on the  books. Underneath it is a fortress because over   six years revenue nearly quadrupled while the  operating margin actually widened from 30% to   43%. With zero debt and more than $12 billion of  net cash, its gross margin looked like it slipped   during the year from 65% to 62%, but for the full  year it came in dead flat at 64%. The same as the   year before, because when low margin cloud orders  ship heavy, the blend dips by design. The only   real catch is the price. Because near 44 times  forward earnings, its peg sits close to a 2. That   happens to be the richest in this group. And the  market made that point when Arista beat and raised   guidance and the stock still fell. So of all the  megas, this is the one where price matters just a   little bit more, which is exactly why I'm looking  for more pullbacks because I want it around that   50 day line where buyers tend to really step back  in. Now we're going to shift and drop down into   the small caps where both the risk and the reward  step up quite a bit. And we're going to start   with Rambus, the shovel seller of the memory  world. Every high speed memory module going   into an AI server needs a small set of traffic  control chips. And Rambus makes the whole set,   led by the conductor chip called the RCD in  what is basically a three company club. So it   gets paid on the memory in the AI build out no  matter which manufacturer is going to win. And   the business underneath is completely transformed  because a decade ago, Rambus was a money losing   patent licensing shell. And today it's a debt free  profitable chip company whose operating margin   swung from a negative 38% to a positive 37%. Now  there's also a twist that runs opposite to what   you would expect. Because Rambus gets paid on the  number of modules that ship, not on the price of   the memory. So when memory goes into shortage and  prices spike, which is fantastic news for Micron,   well fewer modules actually ship and Rambus can  feel that as a headwind. And that is part of why   this is the highest risk name on the list. Right  now it trades near 60 times its earnings with a   peg well above a two. And it's working through  a few overhangs all at once. It is responding   to a document request in a justice department  investigation where it is not charged and it's not   a defendant. A securities inquiry that so far is  just a law firm fishing for plaintiffs rather than   a filed lawsuit. And a chief financial officer who  left earlier this year. Though the company says   there was absolutely no disagreement. So this is a  small high risk position sized down on purpose and   only worth it on a real pullback like the ones  that we've been slowly getting here and there.   Now on the optical side of the build out there's  Fabernet. The purest picks and shovels play in the   whole basket. Because Fabernet is the factory  that physically builds the optical parts that   move data through AI networks. The transceivers  that connect the GPUs and the switches inside a   data center. It builds them for Cisco, Nvidia,  coherent and Lamentum. So it wins no matter   whose brand is on the box. And customers cannot  easily leave because aligning lasers and fibers   to within a fraction of a human hair takes a  lot of years to qualify. And once a program is   certified in Fabernet's clean rooms, moving it is  slow and risky. To me I think the demand is the   real story. Because its data center interconnect  business grew 90% in the past year. And what is   really holding it back isn't the orders, it's  capacity. Since demand is running far ahead of   what it can physically build. Which is exactly why  it's racing to build a giant new plant. Now this   is a thin margin business by nature because the  customer owns the designs. So don't look at the   12% gross margin and expect a heck of a lot more.  The quality shows up elsewhere. Because revenue   more than doubled over 6 years while operating  margin climbed from under 8% to nearly 10%. And it   earns about 20% on its capital with no debt. The  catch is that it's small and it's concentrated.   With Nvidia and Cisco together, about half of its  sales. And even after a 30% pullback, it's not   exactly cheap at a peg of around a 1.5. So this is  the small, higher risk end of the optical trade,   sized accordingly, and a pullback like the ones  we've seen makes the entry work a little bit   easier. So if I'm going to be putting my money  into these today, here's how I'd balance out my   basket. It's going to be heavy on the mega caps,  scaling down through to the small caps. And that's   exactly how I run my own portfolio. And it's  built to be asymmetrical. So the small bets,   if they pay off, great. And if they don't, well,  they don't drag the whole thing down. Well, there   you have it. I hope that you learned something new  today. And as always, thanks so much for watching.