To The Moon

$NVIDIA(NVDA)$  

$Lumentum(LITE)$  

Buying the names the market is selling even on good news.

So from that perspective, we like the semicap guys still, we like (ASML). I mean, their last print they guided for Q4 sales to increase 22 to 30%. And they're expecting that their fiscal twenty-six EUV sales are gonna be 65, targeting 85 and 27, and still have this potential for 110 EUVs in 2028. And so that's what underpins their fiscal 26 sales outlook, which is somewhere between 44 and 45 billion for 36% year over year growth.

And that's against street estimates for 43.1 billion, but the stock still sold off. So those are the kind of opportunities that we're recommending to investors because these are the guys that still have, they're still tethered to come back towards the upside, even on downturns similar to what we saw over the past month.

And that's when we'd also recommend knowing the story, but not necessarily, knowing the story enough to trade the name, not necessarily becoming too attached, buying and holding the stock. A great mentor of mine once put this in a great way. He said you only date stocks, you never fall in love with them. And that's something I think about every day.

Because, for example, when you look at the NeoCloud guys, a lot of these guys have gotten hammered. They're looking at for CoreWeave (CRWV), the worst month since last November and we've seen them shed over 50% from the highs.

CoreWeave, (IREN), Applied (AMAT), (HUT). And this is a group that we've already flagged for investors as higher risk because the business model itself is flawed, right? You have to have this massive upfront capex to keep this business model alive.

Tech becomes obsolete very fast. So you have to continuously have this capital coming in, which really pushes these guys to rely on on external debt markets, et cetera. and so the negatives that the market's pricing in there, I think they're well founded.

At the same time, the prices of these neo clouds is attractive after that dip. And that's what we saw at the end of last week. And that's how we saw them roar back yesterday and on days like today as well.

So I would say for that group where the fundamentals do point to an eventual burst or a shakeout, we think about those guys as trade with caution because we don't see these guys being meaningfully re-rating higher for the longer term.

But we think the market's a bit early in singling this group out because essentially that credit risk that the market's pricing in for the neo clouds, it's much bigger than that neo cloud basket. It's essentially what we think is gonna cause the bigger AI pop.

But to us, the bubble pop is gonna happen in 2027, not this year, because we're still seeing a market that's biting back.

And so when we think about it like that, there's opportunity, there's riskier opportunity, but there's also good fundamental names that have this fundamental basis to come back up. And that's also where, for example, last week we liked the optics names.

So Lumentum (LITE) was one of the names that we triggered an alert on last week when it it dipped, it crashed into its EMA 200. And that's worked out really well. It was one of the top performers yesterday.

Part of the the thesis that underpins that is that at the end of the day, the industry is moving towards NPO and CPO near package optics and co-package optics. And that inherently positions Lumentum and Coherent (COHR) in a good position heading into the second half of 2027.

So with those guys, we like to buy the dip and then we like to trim it to strength and rinse and repeat. Those are the three ways I would say we're thinking about how to handle this kind of volatility. So that's in terms of trading and playing this market.

But I think that what we're getting this earnings season, outside of opportunity to jump into to names with good fundamentals, is we're getting a lot of important data points, I would say.

And I think these are the hints that we need to pinpoint the timing of the bubble come 2027.

When I say that, Tesla's (TSLA) capex cut last week that I think too many people kind of brushed over. I'm not sure if you you saw anyone talk about that or you discussed that.

Basically, this entire cycle, it's only been capex going up, right? Everyone's raising capex every quarter. We're seeing forecasts for what 26, 27 capex is gonna look like, only increasing.

But for the first time last week, we have Tesla cut capex from 25 to 20 billion. And Tesla's not this huge player. We're not talking about Meta, we're not talking about Google (GOOG) (GOOGL), but what it does signal is a shift because this is something that we haven't seen before.

And that happened in the same time that we had Meta also basically hold their capex. They only narrowed it slightly towards the up upside. But basically Meta (META) held their CapEx on last week's call. And if you run a comparison between Meta's Q325 call and Meta's Q326 call, you'll notice that the commentary is much more vague as far as CapEx goes this time around.

And a year ago, the conversation was more about scarcity and needing a lot more compute supply.

This time it's more about how do we get the most efficient kind of compute supply, a lot more focused on unit economics. And this confirms our idea, for starters that yes, the industry is going to shift towards more custom ASIC projects, and we're already seeing that underway.

But it also confirms this idea that it's not only about compute, it's not only about the semi-side of this equation. More and more we need these hyperscalers to deliver on return, and more and more we're seeing them switch up their positioning, if you will, to be able to achieve that.

So even with Meta's Muse, the big focus was okay, yeah, we want to scale Meta Muse. We think that this can be something very big and we can be able to monetize it.

But the main focus was they need to make sure that everything is running efficiently, not only in terms of the capabilities, but also in terms of the compute that's enabling the ability to scale Muse outside of North America as well.

So these are the conversations that are coming out of earnings, and I think they're really important for us to pay attention to because they give us signals on what's to come from 2027. The takeaways thus far is that yeah, ASIC is still we're very much shifting towards ASIC.

We still like Marvell (MRVL) on that front. Yes, the stock has run up, but we found that continuously buying the the dips on Marvell has paid off very well because the fundamentals still back it.

I think come mid-2027, we have to rethink that because of that shift towards NPO and CPO because that's essentially gonna eliminate the DSPs that Marvell sells and is the world's main supplier of.

But thinking now into early 2027, that's still a name that we like because of that ASIC shift and we also like Qualcomm (QCOM) because of that optionality that it can become a very big player in the ASIC scene, in the CPU scene, because one, it has an expertise, it's a legacy player.

And they have the volume capacity at TSM, which has really been a bottleneck for a lot of the players in the industry. So that's in terms of the takeaways we get there. We still do like NVIDIA (NVDA) in the near term.

We don't think it's gonna be an uber outperformer to the SOXX benchmark, but we like it because at the end of the day, they do have this Vera Rubin ramp happening. That's what helped underpin the momentum that we saw.

Even with the sell-off, NVIDIA quarter to date is holding up pretty much in line. I think it was down around seven or eight percent as of last Friday against the SOXX benchmark down about 15%. So we're seeing it hold up, but we're not seeing it be this shining outperformer.

So come 2027, as that ASIC narrative shifts, I think we shift more away from NVIDIA and we grow more cautious there because the industry is going to have to rewrite NVIDIA on this new reality that again, it's sharing the market with the ASIC players. It feels like we've been talking about this for a while, but I don't think we've seen that fully priced in just yet.

Also underpinning this bigger conversation coming out of earnings and also out of the Anthropic S1 is this idea that, what does this shift to more attractive unit economics means for the AI labs?

Because these guys also, similar to what Hock Tan was saying, are still in very much a money losing game and they still need lot of financing or backstops from the likes of Broadcom or the likes of Nvidia.

So when we think about all of these moving parts and we think about Anthropic's gonna be public soon, et cetera, I think that there's a big imbalance in the broader industry right now.

And that's what we expect the market to have to come to to face with in 2027.