In short
How AI capex is surging and shifting from “accelerators” to bottlenecks (optics, memory, networking), and how investors should think about ROI, timing, and mispricings across the AI supply chain and SaaS/cybersecurity.
Guests (backgrounds)
- Amrita (married to Uttam): tech-industry background; covers hyperscalers/software/cybersecurity; publishes on Seeking Alpha (REIT Forum) and Substack; focuses on macro-to-industry demand cycles and ROI.
- Uttam: tech-industry background (Apple/Google; startups); covers “picks and shovels” like semiconductors (accelerators/custom silicon), networking/optics, memory; publishes on Seeking Alpha/Substack.
Key claims (notable examples)
- 2026 hyperscaler capex: Amazon/Meta/Google/Microsoft ~ $680B, ~2x 2025; ROI matters as cash flows weaken and debt funding rises (Microsoft cited down >30% from highs).
- Meta is a top ROI bet: strong AI monetization/margins, forward guidance, less dependence on OpenAI/Anthropic workloads, and expanding custom silicon to reduce NVIDIA GPU reliance.
- Two main AI bottleneck themes: optical AI networking (optical modules/switching; optical CAGR 35–40% 2025–2027; Lumentum/AAOI cited) and memory (HBM/flash; inference-driven demand; memory prices up ~5x in 9 months; Anthropic ARR cited rising from ~$5B to ~$30B).
- Mispricing in software/cybersecurity: software derated (Monday.com, Wix, Atlassian mentioned); cybersecurity viewed as mispriced and “full stack” (CrowdStrike, Cloudflare). Anthropic “Project Glasswing” cited as evidence AI won’t kill cybersecurity and supports platform consolidation.
- ServiceNow risk: valuation cheap, but market wants faster revenue acceleration; Anthropic managed agents seen as a competitive pressure.
- NVIDIA valuation/fundamentals divergence: smaller beats vs prior years; China revenue uncertainty; investors want clarity on $62–65B cash use (buybacks/dividends vs supply-chain commitments).
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Chapters
Tap a time to open that second in VOGetting to Know the Guests
0:45 to 2:52
The hosts engage with the guests about their backgrounds and experience in investing.
“So I collaborate with Colorado Wealth Management on his investing, though I do not actively contribute because most of my analysis is mostly on tech.”
The Impact of COVID on Investing
2:52 to 4:18
Discussion on how the macroeconomic environment during COVID influenced their investment strategies.
“And that's how our investing sort of journey started.”
Navigating the AI Investment Landscape
4:18 to 7:17
The guests share their thoughts on AI investment strategies and the rapidly changing tech landscape.
“Well, first of all, it sounds as I'm hearing you both talk, it sounds a little bit like when I started at Seeking Alpha was in 2008 and the great financial crisis.”
Analyzing the AI Supply Chain
7:17 to 10:04
The guests explain their approaches to analyzing the AI supply chain and identifying investment opportunities.
“And, you know, you would have already seen this kind of research in, you know, both our research on the platform.”
CapEx Trends in Hyperscalers
10:04 to 14:06
Discussion on the projected CapEx by major hyperscalers and its implications for the market.
“Do you guys each analyze the same data and then come together and discuss it?”
AI CapEx and Meta's Position
14:06 to 16:44
Learn how companies like Meta are leading in AI monetization and CapEx ROI.
“And the companies that can or the company or the companies that can actually demonstrate the quicker or the faster conversion of this CapEx into meaningful revenue acceleration or margin acceleration will eventually win.”
Shifts in AI Infrastructure: Optical and Memory Trends
16:44 to 21:06
Discover the critical shifts in AI networking towards optical solutions and memory demand.
“So, yeah, so ROI on CapEx is one of the major themes.”
The Importance of Gross Margin Expansion
21:06 to 28:00
Understand why gross margin expansion is crucial for companies in the AI space.
“And they were only able to do it because AI inference has exploded, which has created this massive demand for memory.”
The Importance of Gross Margin Expansion in AI Companies
28:00 to 36:50
Learn how gross margin expansion is crucial for companies to thrive in the AI market.
“how does a company carry on business as usual?”
Mispricings in the Software and Cybersecurity Sectors
36:50 to 42:05
Explore the current mispricings in software and cybersecurity companies amidst AI disruption.
“Amrita, you mentioned mispricings, looking for mispricings at the start of the conversation.”
Show all 13 chapters
Cybersecurity and Software Valuations
42:05 to 46:09
Explore the relationship between AI advancements and the current state of cybersecurity companies.
“So I think cybersecurity is one of the probably the safest spots in SaaS right now or in the entire AI eating SaaS or AI eating cybersecurity and the whole software space in general.”
Market Risks and Meta's Position
46:10 to 48:35
Understand the macroeconomic factors affecting Meta and the broader market.
“given its data advantage and enterprise customer relations.”
Investment Strategies and Metrics
48:36 to 55:06
Learn about key metrics and strategies for navigating current market volatility.
“I mean, yes, there are other risks like regulation and things like that.”
Transcript
Automatic transcript. May contain errors.0:00Thank you.
0:30Uttam Dey:Vaughn called you his favorite analyst on Seeking Alpha. I don't think I'm overstating that. I think that's what he called. So yeah, lots of your reputations precede you. Amrita, you have an investing group also on Seeking Alpha called the REIT Forum. Is that correct? That's correct. So I collaborate with Colorado Wealth Management on his investing, though I do not actively contribute because most of my analysis is mostly on tech. But yes, that is correct. Okay. Wanted to make sure about that. So maybe if you would both introduce yourselves and your strategy. Did I mention that you're married?
1:07Uttam Dey:That might be our first time having a married couple on the show. So, yeah, we are married and yeah, we have been married a very long time, probably eight years together for probably what, 11 or 12 years. But yes, we work very well together and it was always our dream to sort of do this thing where we each have our individual strengths to bring together, especially investing. And curious, did investing bring you guys together? Did that have anything to do with you meeting or? That would have been cute. But no, actually, we met in university and both of us have quite extensive backgrounds working in technology companies in the Bay Area.
1:51Uttam Dey:So Uttam actually worked in a larger sort of well-established companies like Apple and Google. And I worked in various startups that eventually some of them got sold to larger companies. So, yeah, so we do come from a very deep technology background. And I think probably around like 2018, 2019, when it was a Robinhood picking up steam in like the local conversations, we were probably one of the early users and sort of really got into this active investing phase in our lives. And probably around 2020, 2021, I think both of us wanted a career pivot where we just wanted to understand more about just core investing in general and just really connect the dots, understand innovation cycles, how each company sort of fit into those cycles, how to sort of understand balance sheets, understand financial statements to really understand how demand cycles and margins are about to inflect higher or lower.
2:51Uttam Dey:And how that dictates price actions on various stocks. And sort of we got deep into that. And that's how our investing sort of journey started. We started publishing on Seeking Alpha. We also have our own Substack group where we publish our research. And yeah, it's been, what, two years that we're doing it full time and absolutely loving it. Did that, I mean, I see that that coincided with COVID. did that macro picture lend itself to that decision? Absolutely, yes. I think it was probably the first time where I took an active interest in macro because I think prior to that, I didn't really pay much attention to macro at all.
3:32Uttam Dey:But yeah, it was probably the first time that I really wanted to understand and connect the dots in how sort of global financial economies work.
3:41Amrita Roy:And then, you know, in addition to retail going macro, So my focus would always be kind of going towards the industry side and trying to understand which industries are moving up and down in the various demand cycles and trying to get deep inside the industry as well, getting that subject matter expertise there. So I think like those two different kinds of research methodologies side to like just amalgamate and that, you know, that's how this research started moving ahead towards investments. sense?
4:18Uttam Dey:Well, first of all, it sounds as I'm hearing you both talk, it sounds a little bit like when I started at Seeking Alpha was in 2008 and the great financial crisis. And I felt like that was the number one thing that I was taking away from my time at Seeking Alpha, which was connecting these dots and seeing this macro picture that felt very scary and something that felt very confusing and what the heck is going on. And if you start connecting the dots, it may not make you feel better, but it makes a lot more sense. I know. So I feel like that's, yeah, that's, and when you're talking about the tech space, you probably couldn't have, I mean, maybe you could talk about different times in history, but these past two years in the tech space are supremely interesting.
5:01Uttam Dey:And, uh, I bet it lends itself to a lot of thought provoking conversations between the two of you. So maybe ground us how you're thinking about maybe how you've been thinking about the past couple of years and then bring us up to date with how you're thinking about today. I'll let you decide who goes first. You guys probably haven't worked out better than I do. I could. Yeah, you go ahead. Yeah, and I'll fill in the dots.
5:28Amrita Roy:As always. Generally speaking, I think, you know, we would focus on just investing in high growth assets, you know, the kind that tend to generate strong alpha that generally sit in AI or the digitally native ecosystem. So that's like, you know, what the market or the investment industry just classifies as the technology sector. And, you know, we just spoke about technology a lot and how that happens to be our background and, you know, just running in our veins. So we invest in our areas of expertise. But, you know, to the point that you just mentioned earlier, you know, these last two years have been just...
6:10Amrita Roy:crazy for technology. So off late, what we've started doing is we've started utilizing a lot of thematic investment strategies to deploy our capital, possibly even in multi-phased approaches because let's face it, this is the AI era. And while that's exciting, there's lots of extremely fast movements in terms of, you know, AI innovation, capital investments, R &D. There's lots of competitive positioning also that's, you know, quickly moving and it's really rapid. You know, in one quarter of the industry, we're talking about like, you know, investing in picks and shovels. In the other quarter of the industry, we're talking about, you know, where the memory shortages or, you know, nowadays the word on everyone's lips is bottlenecks, right?
7:02Amrita Roy:So the The point is, in AI, it's just fast and furious, and AI is moving really fast. And with that pace in mind, portfolio capital is following rapidly as well. And so we've adjusted our portfolio strategy as well to not just be secular, but also have some kind of thematic investment to accommodate for those kinds of strategies as well. I think like you know on a more on a more tactical basis I would say like we tend to go top down a lot so we look at industries on the macro level we look at dependencies competitive positioning and then you know try to move down through the industry and you know pick out winners winners which will they'll they'll deliver these category leading performances over the rest of its peers.
7:56Amrita Roy:And, you know, you would have already seen this kind of research in, you know, both our research on the platform. You know, when it comes to capital deployment, finally, I think that's the last bit I just wanted to say, like when it comes to capital deployment, we usually just, you know, start layering into a position. You know, we start underweight. And then as our conviction with the name starts getting, you know, a lot better, we'll build our exposure to full weight or even overweight, depending on, you know, how our conviction flows. But then, you know, when it comes to, let's say, thematic investing, we might just go really aggressive on a particular name.
8:35Amrita Roy:But that's all based on how, you know, our risk and our portfolio management strategies fit in.
8:42Uttam Dey:Yeah, I mean, I think just to summarize, really, like, you know, like AI is not a constant, right? It's like over the last two years, we have seen leadership change, shift from AI accelerators to now in the bottleneck trades, such as memory, optics, networking and things like that. In the meantime, we have also seen certain corners of the industry, which used to generate alpha like software, completely get disrupted. As well as cybersecurity, which is currently sort of going through this really messy stage in between. so I think a big chunk of our job really is to just understand where these bottlenecks are and where the next bottleneck could be and at the same time just maybe a lot of the time it's also about identifying where a lot of the mispricings could be as well where demand may be about to inflect but the market has no idea or the market is not pricing in that the demand is about to inflect higher and it's a constant process and I think yeah and given the the pace of AI progress, which is so much faster than like, let's say, cloud computing or the previous innovation waves.
9:53Uttam Dey:This is sort of like a quarterly thing that we have quarterly or even like monthly thing that we have to be constantly updating ourselves on to remain on the right side of the trade, essentially. What's your process like in terms of are you each interested? Do you guys each analyze the same data and then come together and discuss it? Are you each analyzing different points, different stocks. How do you guys blend your processes or work together? Sure. And also just curious, how do you decide like which articles you write, which stocks that you each write about? Sure, I can go for that. I think in terms of the AI supply chain, if you look at it, I cover most of things that sort of the top of the AI supply chains are things like hyperscalers, software, cybersecurity.
10:40Uttam Dey:Those are my sort of expert areas whereas Uttam covers more of the picks and shovels like semiconductors, so accelerators, custom XPUs, networking, optics, memory, and things like that. So our stocks rarely, the kind of research we publish on Seeking Alpha, the stocks never collide with each other given our individual sort of expertise, expert areas.
11:05Amrita Roy:But they're all part of the same AI ecosystem.
11:07Uttam Dey:Yeah, they're all part of the same AI ecosystem. Yeah. And so, yeah, I do my own research. But at the same time, it's the same top-down research that both of us perform on our individual sectors, industries, stocks within. We obviously discuss in terms of our overall portfolio strategy, how do we go about, given sort of the risk-reward we are seeing across the stocks that we have identified as potential winners. And based on that, we decide on capital allocation, price points, entry exit strategies, etc. OK, so in terms of thematic investing, what would you say are the top two themes that you're thinking about and what stocks are doing in those subsectors accordingly?
11:54Uttam Dey:Again, I think I will start at a high level because I think and then you can sort of and I will point out one core theme within and then you can talk about your themes in the stocks. so at a high level when I look at AI I think one of the biggest signals we have right now is AI capex from the big hyperscalers because that is ultimately what is driving the demand cycle in the entire chain if you like across your AI accelerators manufacturing equipment memory you name it. And for 2026, the four big hyperscalers, which include Amazon, Meta, Google, and Microsoft, have committed to spend approximately$680 billion in CapEx, which is far higher than what analysts were originally expecting, and is almost double that of 2025 numbers.
12:53Uttam Dey:And all this money is essentially going into building data centers, securing GPUs or advancing their own custom silicon roadmaps and or securing power contracts and things like that. But the thing that is unique for 2026 is that we are finally reaching a point where up until now, most of this capex was actually being driven by these hyperskillers using their own operating cash flows. But since revenue is not necessarily catching up yet at the speed, these companies are now going to have to increasingly tap into the debt markets to fund their cash, to fund their capex, essentially. And this is why probably we are seeing the hyperscalers or the heavyweights in the S &P 500 pretty much stall in performance.
13:45Uttam Dey:Microsoft has had a pretty horrendous decline of more than 30 % from its all-time highs. as investors are like, okay, so like free cash flow is going to go negative right now. So how do we sort of account for that risk? So I think ROI on AI CapEx is one of the big themes that is going to be on investors' mind in 2026 and as well as in the coming years. And the companies that can or the company or the companies that can actually demonstrate the quicker or the faster conversion of this CapEx into meaningful revenue acceleration or margin acceleration will eventually win. And I think in that respect, I mean, I have written about it on Sigling Alpha platform and we are huge investors in the company as well.
14:31It's our second largest position in the portfolio,
14:34Uttam Dey:which is meta. We particularly are quite bullish on meta given the fact that I believe, that we both of us believe that it's AI monetization is one of the strongest amongst all hyperscalers. Like we can see that in the revenue acceleration, margin acceleration the company has demonstrated. It's forward guidance, which also calls for acceleration. Its revenue is not dependent on things like OpenAI or Anthropic actually securing workloads. There is no customer concentration issues as such. So it's pretty much sort of shielded from all these customer concentration issues that companies like Microsoft, Oracle are facing.
15:13Uttam Dey:And then, yeah, just generally the work it is doing in its custom silicon, I think last month it expanded its portfolio of new custom silicon that is due to come, which will reduce their dependence on NVIDIA GPUs. It's a huge boost to margins, which I don't think markets are pricing in. And again, yesterday we heard something on the Muse AI model that they have made some progress on. But again, like I have said in my posts as well, like I don't think that Meta needs to win at AI model race because it has the distribution power. It has the it has sort of the rails to serve AI at scale. So I think, yeah, like I think both of us are quite bullish on Meta, given that we have given the fact that we believe that it has the best chance at proving the ROI of the ROI of AI CapEx at the fastest rate.
16:12Uttam Dey:And I also think that Amazon and Google are also pretty well positioned as well. I mean, I think increasingly more so, I think both of us are of the opinion that the faster these companies can can can sort of vertically integrate with their own custom silicon. Obviously, Google with its TPUs, Amazon with its Tranium and Graviton. I think it's just a matter of time that investors will slowly start to believe in these companies' new revenue sort of angle with their custom silicon, essentially. So, yeah, so ROI on CapEx is one of the major themes. And meta plays a major role, along with Amazon and Google, that probably are lesser rates than our meta conviction.
16:54Uttam Dey:but yeah that's that's one of the themes i wanted to discuss as amrita does typically she starts
17:01Amrita Roy:like really macro and then i started joining you know from an industry standpoint um in in my from my point of view i think there are in for example if you just asked me this question like a year ago i would have just said uh ai networking great uh that was a great theme last year and probably still a good theme this year as well. So last year, for example, Celestica. Celestica did really well as like the standout star in AI networking. And I know Steven Kress is probably going to be really happy that I took Celestica's name. But yeah, that was last year. Today, I think that there are just two key areas.
17:43Amrita Roy:The first is, you know, this whole optical AI play. And then the other is memory. And because for a simple reason, because in my point of view, I can see that there is a direct cause and effect that can be established between AI inference and the shockingly catalytic and disruptive impact that AI inference has had on the AI economy in these past nine to 12 months. So most importantly, there has been no part of the AI ecosystem that is more impacted by AI inference than scale-up networks. And for the simple reason, like last year, if you saw scale-up networks, the whole industry was talking about getting together 100 ,000 GPUs to be connected on a single fabric, which could then run as one single compute instance.
18:42Amrita Roy:and now there is definitive talk about adding 1 million GPUs together on the same single network fabric, that's a scale-up network, so that it can run as one single compute instance. What that does is in the last 15 months, you're moving from 100 ,000 GPUs to 1 million GPUs. You're significantly raising the complexity that is now being engineered in the AI network. There's a lot of load, a lot of bandwidth issues that could be created. and the existing network infrastructure is kind of legacy because you're moving from that scale. So you're now, the industry is thinking outside the box, how to support this network still and get that performance.
19:25Amrita Roy:And that's how the industry gravitated towards optical networks. So we're talking about adding laser chips. We're talking about adding optical modules. Modules, we're talking about switching out these legacy switches that were there and adding laser within the networking switches. So there are all these new optical products that's coming in, which is why you could see how there's this rapid demand for optical products to be added inside the AI networking infrastructure, which we can see today. And mind you, the demand is so intense, which I recently wrote about it. I wrote my five top stocks in the optical AI space.
20:14Amrita Roy:What I was saying before that was the demand is so intense that, remember, the entire industry, the entire optical AI industry is moving at a pace of 35 % to 40 % CAGR between 2025 and 2027. Remember, again, the whole industry is moving at such a rapid pace, which is incredible. And then within that, you have small players like, you know, you have Lumentum, you have AAOI, that's Applied Opto Electronics, which is, they're probably growing at, you know, 70%, 100 % year on year. So these are the kind of companies that, you know, you really want to invest in. probably and i just you know talked about optics because probably i i feel like the ai industry had probably sort of planned for for this switch from the legacy network infrastructure to optics what they did not plan for is the second theme i'm going to talk about that's a memory um you know in in the past nine to twelve months um ai ai inference you know the the the the demand vector that that ai inference created was massive because there was this whole move um to to monetize ai agentic ai came in um ai models started getting really better and you could also you could see like how how much the demand is at by at the end user because uh startups like anthropic for example last year arr 5 billion recently in the last few days arr 30 billion 6x so that's that's the demand that you're seeing from the end user in companies like Anthropic.
21:55Amrita Roy:And they were only able to do it because AI inference has exploded, which has created this massive demand for memory. Memory like you have high bandwidth HBM, high bandwidth memory, flash memory that's being added in scale-up networks. The prices for these memory products have skyrocketed about five to just in the last nine months. So that's created this huge other demand vector for memory as well. So I think from an industry standpoint, answering your question about thematic investing, optics and memory, my two top two picks for 2026.
22:43Uttam Dey:Well, you brought up Steve Kress and Lumentum was one of his top AI picks for the year. And he was just on yesterday talking about how it's one of his strong buys still, even with the recent run up. I'm curious, A, do either of you, do both of you use the quant system at all, use like the quant factor grades at all? And B, anything else that you would add to the Lumentum conversation specifically? And then if I may tag on, since NVIDIA is also part of this conversation, something else we got into was the valuation. I know that that's a lot at once, but they kind of may all have to do with the same core point.
23:23Amrita Roy:I can talk actually about both about NVIDIA and about, I think, Lumentum and about maybe...
23:31Uttam Dey:I think your question was also about whether we use the Quant tool. The Quant tool, correct? Yeah, yeah. Yeah, I use it quite a lot in order to, and I add a few more layers of my own analysis to decide. Again, I think, as I said, like a lot of our work is to understand which companies are at an inflection point. And I think the Quant tool is one of the tools that I definitely use to assess certain companies. Again, I do not cover memory and optics, so I'm not going to talk on your behalf.
24:04Amrita Roy:I think Steve will be really happy to hear this.
24:07Uttam Dey:Um, so when I was fishing, by the way, I'm not fishing all honest answers. No, no fishing here. Yeah, please. No,
24:17Amrita Roy:I had been researching about AI networks. So the first time I started getting convinced about AI networks is when I heard in December 2024, I heard Hawk Tan, he's a CEO of Broadcom. And he started talking about how AI networks is really underpriced. There's a big boom that's coming and the industry has not priced in the big boom that's coming. So I started doing my research on AI network right from December 2024. I still hadn't started investing in the AI network space. I was probably looking at Arista networks at the time. And then you had this whole liberation day tariffs, volatility that had kicked in around that time.
25:02Amrita Roy:And I did come across Celestica and I did see, wait, there's a strong buy rating, you know, which I saw on Seeking Alpha. My conviction at the time did raise, did get raised. We started discussing this, you know, in that month itself. And I think in April or May itself, we started, you know, putting in a lot of our capital towards Celestica. So Seeking Alpha's quant rating mechanism, I think it did play some role in our capital deployment at the time. And Celestica is kind of like the anecdote that can explain that.
Read the full transcript
25:42Uttam Dey:And then the NVIDIA Lumentum point.
25:46Amrita Roy:Yeah.
25:47Uttam Dey:You forgot about that. Just moved on. That's what I'm here for.
25:54Amrita Roy:about Lumentum I I'm quite bullish on Lumentum also it's it's in in in my um in the in the recent post that I put out the top five picks for um the AI optical revolution I it just got published a few days ago I I believe Lumentum is a great pick as well so I in that pick in that list I have Lumentum, Coherent Applied Opto Electronics, Marvel, and then Sienna. But I think of all of them, if you had to just ask me to narrow it down, I would definitely say Lumentum and Marvel maybe, along with Applied Opto Electronics. These are my three further narrowed down picks. With Lumentum, what's really great about them is they're delivering 70 % plus revenue today.
26:46Amrita Roy:They're also demonstrating this huge, massive gross margin expansion. By the way, I love watching gross margin expansion in 2026. I think it's apart from the ROI metric that Amrita watches at the macro level, at the industry level, I just love in 2026, love watching when companies can demonstrate gross margin expansion. And Lumentum is one of those companies that's able to do that.
27:12Uttam Dey:Sorry, can you just explain why gross margins are so important for you?
27:16Amrita Roy:The simple reason is that, A, you have these really fast-paced investments from the capital side, capital investments, that's moving into AI. Innovation is moving at such a fast pace, like I just said. And then the other, on the macro side, you have these big geopolitical tension risks. You have inflation risks that's now starting to get priced in. You have supply shocks that's starting to get priced in. Probably a few weeks ago, people were talking about helium shortages. There's obviously memory shortages. So with all of these shortages, supply shocks and inflation risks, and as well as geopolitical tension, how does a company carry on business as usual?
28:06Amrita Roy:not just by growing their revenues at a 50, 60, 70 % because AI demand is there, but actually by demonstrating that your unit economics of the business is sound. The only way to show that despite all these supply shocks and inflation risks is to demonstrate gross margin expansion coupled with top-line growth. If you can do that, the company will be rewarded immensely in 2026. But if you can't do that, no matter whether you're NVIDIA or even Broadcom, for example, you're not going to be rewarded accordingly. And Lumentum is doing that. So Lumentum, apart from the 70 % plus growth that they are already on the path to deliver, they're also saying they're able to, I think, if I remember correctly, about 65 % or 70 % next year as well, or at least that's what the analysts are projecting.
29:00Amrita Roy:But Lumentum is already delivering gross margin expansion. So I remember doing this in one of my posts. I did this comparison between Coherent, that's growing gross margins, as well as Lumentum. And you see this rapid, significant gross margin expansion from probably about, I don't remember the bottom, but it's probably about like 15, 20 percent, all the way to 30, 35 percent. and all these three companies, Lumentum, Coherent and Applied Opto Electronics, they're all trying to target 40 % gross margin numbers by the end of the year. So like I said, high growth, rapid gross margin expansion and Lumentum is able to do that already.
29:44Amrita Roy:What I like about Applied Opto Electronics, by the way, in addition to all of this, is that they own the entire manufacturing ecosystem all the way right from like really manufacturing that laser chip the ingredients that go into that like for example indium indium phosphide is one of the core ingredients so applied optoelectronics is able to manufacture everything all the way from you know the little laser chip to the the optical plug the optical pluggables that um hyperscalers use to do for their for their connectivity applied up to electronics is doing all of that. So they're in command of their supply chain.
30:26Amrita Roy:They're able, they're already deploying CapEx to ramp up that capacity. So that's another good company as well. And then I think your last question was about NVIDIA. Was that about?
30:37Uttam Dey:Well, I was going to say this notion of that the valuation and the pricing and the fundamentals are a bit divergent these days and how kind of to think about that.
30:50Amrita Roy:So about NVIDIA, NVIDIA, it's in a peculiar spot right now. On one hand, they're doing everything, I think, that I just spoke about. They're growing their revenues, especially their data center revenues. And at GTC, with all of these product announcements, Jensen Huang has confidently echoed this outlook where he can navigate the company towards accelerating revenues. And by the way, he's already showed that in Q4 of 2025. They're also signaling gross margin expansion. So that's the other plus point. But what's peculiar about NVIDIA since last year is there are two things that make NVIDIA stand out, I think, if I remember correctly.
31:47Amrita Roy:The first is there's no company which is a bigger beneficiary of AI than NVIDIA. through 2023 and 2024, NVIDIA was, I'm not even going to talk about, you know, how the revenues were growing. I'm just going to talk about the magnitude of beat that NVIDIA was delivering. They were delivering, you know, earning beats, revenue beats of like 5%, 10%. Right now, all that they're doing is they're delivering an earnings beat or a revenue beat of just 2%, 3%, maybe single digit, low single digits. that's not something investors were probably used to by the way if you look at q4 2025 their china revenues used to be between 4 billion and 6 billion on on you know per quarter in q4 2025 their china revenues grew to about 8 billion so now all through when you start adding that up.
32:54Amrita Roy:You have a company that's delivering revenue beats of about 1 % to 2%. You also have the company where they suddenly delivered China revenues of about$8 billion in Q4 2025. But the same management last year was saying, no, you should exclude China revenues from the expectations. But now that China revenues have come in, which is double, they're able to deliver that little bit of an earnings or a revenue beat. That is probably some kind of a problem that investors have where they're unable to understand now, how do I price the forward revenue outlook? Is it going to be with China? Is it going to be without China?
33:41Amrita Roy:Is China revenues going to expand faster or is it not? Because there's still lots of dependencies, there's still lots of uncertainty about trade restrictions, licenses. How do investors price that forward? But even if an investor wants to ignore all of that aside, the main factor, I think, for NVIDIA stock is they hold about, I think, $62 or$65 billion worth of cash on their balance sheets. The big question now for every investor is, what is NVIDIA going to do with that$60 to$65 billion worth of cash that's there on their balance sheets? You have, over the past two years, you have a huge slate of institutional investors that have come in.
34:31Amrita Roy:They're holding probably about 70 % of, 60-70 % of NVIDIA's stock base. And all these investors, especially these institutional investors, they're starting to ask for what a typical institutional investor would ask for. They're starting to ask for shareholder value, at least, you know, increasing shareholder value. They want that cash. They want a piece of that cash. You could give me, as an institutional investor, you could give that to me in terms of dividend. You could give that to me in terms of buyback. But I need a piece of that cash. What NVIDIA's management is saying, however, is that we're going to prioritize first growth.
35:14Amrita Roy:or rather we're going to prioritize securing our supply chain commitments, which actually is quite, it's actually right. You know, as a growth investor like me, I think they're doing something right because if you don't secure your supply chain right now in the face of all these geopolitical tensions and supply shocks, you look at, you know, what's happening in the memory space, NVIDIA is going to have margin pressures down the line. So if NVIDIA's management right today is able to prioritize its cash towards securing supply chain commitments, I think it's right. But an institutional investor, they're more concerned about buybacks and dividends.
36:01Amrita Roy:And there's this disconnect that's happening between these two cohorts of the investor, which is why if you just look at NVIDIA stock over the past eight or nine months. It's a rattling between like 160, 180. There's this immense battle that's going on. And I think until NVIDIA demonstrates
36:26Amrita Roy:either this significant revenue growth where they can show 5 % plus earnings beat or revenue beat, or they can say that, okay, now we're going to start prioritizing our cash towards dividends or buybacks, you're going to see some movement in the stock. I think that's what our point of view is.
36:50Uttam Dey:Amrita, you mentioned mispricings, looking for mispricings at the start of the conversation. Anything else to add to the mispricing conversation? Like which names you're looking, which names come to mind? Sure. Yeah, so I would like to add something to the mispricing piece of the conversation. And I think, like I said at the start, like, you know, AI, in the AI trade, there are obviously the bottleneck trade right now that is absolutely outperforming everything. Like companies in the optics and memory space do not take a breather any single day, like pretty much like they're just making higher highs every single day.
37:26Uttam Dey:And yet you have pockets in software as well as in cybersecurity, which are getting no love. And there is a lot of conversation about especially software getting disrupted by AI. And I would like to add the mispricing part to the AI eating SaaS conversation. So, yes, I think AI does have a big role to play in disrupting parts of SaaS right now. So we are seeing companies like Monday.com, Wix, perhaps even Atlassian, and generally the entire software sector has gotten immensely derated, like to a point which we have not seen, I don't know when in history, But it has been completely massacred, this space.
38:20Uttam Dey:The question obviously lies as to, is there going to be a revamp in software in general? And then just this morning, I think I was reading something around Anthropic launching something called managed agents, where now it's going to be much easier for developers to just develop and deploy AI agents, which was not previously possible. This is, again, like a space that ServiceNow has a strong hold in. So despite me being bullish on ServiceNow, I think I need to take another look at ServiceNow right now. But coming back to mispricing, I think one of the micro sectors within SaaS or software as a service that is getting also clubbed into this whole massacre is cybersecurity.
39:09Uttam Dey:And I think cybersecurity is one of the areas which is currently very, very mispriced. So far, I think since the start of the year, we have had announcements from Anthropic regarding their cloud code security, which just, you know, hit the cybersecurity complex like a rock. I wrote a post about it explaining like why it like how why the fears were irrational. And during that time, we also doubled down on our position and CrowdStrike and Cloudflare. And soon afterwards, the sentiment did improve because of the war in Iran, where a lot of investors flocked into cybersecurity stocks as a form of hedge.
39:52Uttam Dey:But then again, I think with the series of cybersecurity incidents that happened at the end of March, starting with Anthropics, Cloud Code source leak, Mythos leak, along with the Mercos supply chain attack. I think the whole narrative in cybersecurity is very, very sort of volatile at the moment. But I think, I don't think when it comes to cybersecurity, it should not be viewed as a standalone solution. It is a full stack solution. And the biggest evidence of that came yesterday or two days ago from Anthropics Glasswing announcement, Project Glasswing Initiative, where they sort of created a security coalition with CrowdStrike, Palo Alto, and 40 other companies, plus some more, to really sort of like get together and prepare for what is to come as AI models are becoming more and more capable.
40:58Uttam Dey:And I think on that front, particularly, I think CrowdStrike is one of the biggest mispriced stocks at the moment. Because it has a full stack solution with its Falcon platform. It has cloud security. It has next gen identity solutions. At the same time, it has also got partnerships with Cloudflare and Zscaler for zero trust solutions. So really, it's giving proper competition to companies like Palo Alto Networks and Fortinet, which has been actually sort of, you know, has been put in the garbage bag right now. But coming back to CrowdStrike, I think it's one of the biggest mispriced talks because one of the biggest things we are also noticing in cybersecurity is that there's a lot of vendor consolidation that is happening.
41:46Uttam Dey:Enterprises are moving away from like point solutions and consolidating in like full stack solutions like CrowdStrike or even Palo Alto. So like, you know, you would often hear words like platformization, for instance. So companies that have a full stack solution offering a full spectrum of cybersecurity solutions. And now that is validated by Anthropics Project Glasswing that AI won't kill cybersecurity, but rather these companies are essential partners to like power the next generation of cybersecurity solutions as AI becomes more capable is something that the market hasn't woken up yet. So I think cybersecurity is one of the probably the safest spots in SaaS right now or in the entire AI eating SaaS or AI eating cybersecurity and the whole software space in general.
42:37Uttam Dey:I would say CrowdStrike, Cloudflare, particularly CrowdStrike is one of my biggest and favorite picks. Appreciate that. I'm curious. You mentioned service now that you may have to rethink your take there. What does that look like? Like, A, what makes you rethink something? Is it a thematic? Is it narrative driven? Is it fundamental? And then how do you go about rethinking your strategy? I think, well, it is fundamental. I think at a valuation level, these companies have never been cheaper. So if I go with the valuation argument, these are like, you know, bargain hunters, bargain stocks. But I think this is not the way to think about a sector which the whole industry or where everyone is thinking that is about to be disrupted.
43:27Uttam Dey:so yeah from a valuation perspective I have no doubt that the entire software sector is trading at a discount including ServiceNow I think when it comes to ServiceNow over the last in the last quarter you know the company demonstrated RPO acceleration its margins are stable its retention rates are strong it has also been making some solid progress with its sort of AI control tower, where now assist AI solution is part of. This is part of the AI agentic roadmap that it has been building over the last 12 months or even more, perhaps. But the thing is that it has not been able to move the needle on revenue acceleration.
44:14Uttam Dey:And what investors really, really want to see is like movement on revenue. what investors really want to see is revenue acceleration, not 20 % revenue this year, next year, the following year. They do not want to see like, because 20 % revenue continuously over a three-year period projections is quite strong, but not anymore in this agentic AI era, I think. I think in order for these companies to be re-rated higher, investors want to see these companies actually be able to accelerate their revenues. So the company, I think...
44:50Amrita Roy:That becomes really tough when you have Anthropic. The point that I made earlier, you have Anthropic that they were probably on the pace of generating about 4 to 5 billion in ARR last year, which is significantly scaled to 30 billion just a few days ago in one of these announcements. So if Anthropic can generate that kind of significant acceleration, the market's now looking at ServiceNow and saying, we don't just want 20-25 % stabilization in the revenue growth. We want an acceleration right now. And there needs to be some meaningful acceleration from here on. And that's not happening yet. Yeah.
45:36Uttam Dey:So, yeah. So, so with the managed agents piece announcement from Anthropic, it'll be curious to see how, you know, how, how, how the earnings call goes later in the month, this month. Like I said, I think, yeah, it's not about just beating the current earnings and revenue estimates, which I'm, I'm quite confident service now will. but it's about really showcasing that it has, you know, it's usage-based revenue. It's now Assist can be a real contender to Anthropik's solution or that Anthropik really doesn't pose that big of a threat given its data advantage and enterprise customer relations. And so far, markets are not convinced and I don't think the management yet has the numbers to showcase that.
46:21Uttam Dey:And until that happens, yeah, the entire software space will be kind of depressed. And like I said, I think in the entire depression of the software space, cybersecurity is caught, unfortunately. And that piece I do not believe is, and that is a mispricing part that I talked about earlier, which where cybersecurity actually should be benefiting from all of this. But right now markets are just like, shoot now, think later mode, really. Amrita, I also wanted to ask you about meta. You were bullish on them. I'm curious if you have any compelling risks in the back or the front of your mind. Sure. I think the biggest risk to meta at this point is really the macro backdrop, to be very honest.
47:13Uttam Dey:And I can talk about this even from a portfolio management standpoint, because like yesterday and even perhaps even today, we are seeing the markets bounce higher. but we are but is this bounce really sort of a tradable bounce or a durable bounce and i think that until we get some sort of a confirmation that inflation has peaked the higher for longer narrative will stick in the market and as long as the higher for longer narrative sticks in the market equities will be pressured because you know higher for longer generally elevates the risk premium and equities get pressured as a result. So I think from a general market direction standpoint, we do expect inflation to tick up higher in the short term.
48:05Uttam Dey:And as a result, we do consider the current bounce in the S &P 500 as sort of like a tradable bounce before we probably see more weakness in the summer. And I think with every S &P 500 weakness, We will obviously see more pressure building up in each of these stocks. And that is not because necessarily of like earnings, downward revision and earnings, but because multiples are getting compressed. And I think that is probably the biggest risk that I see for Meta at this point. I mean, yes, there are other risks like regulation and things like that. But I feel like those are just story. I mean, those are past us.
48:44Uttam Dey:I think those are not fundamental structural issues that can harm the long-term path of a stock. But from a position building standpoint in a company, these are the sort of short-term volatility that needs to be taken into mind, needs to be taken into account from a positioning portfolio allocation cost basis standpoint. There's so many more questions that I can ask you. So I hope that this is just the first time that you both come on. And I hope this is just the first conversation we have because there are so many more names and things to get into. But as we wind down this conversation, I'll ask you both, Udam, I'll start with you.
49:25Uttam Dey:A, what are you thinking about in the coming days, the coming weeks, the coming months? Yeah, I guess I'll just leave it there. Anything you want to put in there?
49:32Amrita Roy:My thoughts out here would just be an echo of the tactical sort of outlook that Amrita had just mentioned a short while ago that we do expect some kind of a tradable bounce that's going to keep moving the markets higher. there could be some more volatility that we might see through the next few months, maybe into the summer. And we do then expect some, this volatility to then just sort of find some resolution. But through all of that, I think for investors, it's just a matter of keeping calm, watching a few metrics. That's really important. Not many, just a few. And then moving ahead with their portfolio capital deployment strategies just as they are comfortable with their risk appetite.
50:30Uttam Dey:Udam, what would you add? I know you mentioned gross margins earlier, but what would you add to the few metrics? What would you consider the most salient for your average retail investor?
50:39Amrita Roy:For semiconductor companies especially, whether it's memory, GPUs, CPUs, memory, equipment, it doesn't matter. I think gross margins rules all of the metrics. One of the other metrics that I watch a lot is the amount of inventory that a company holds on its balance sheet. And that sort of tells me how the gross margins could be moving ahead. The more the inventory, the more pressure there could be on the gross margins. It also sometimes would signal that there is going to be some more pressure that's coming in in the next quarter or two. Sometimes, though, the companies do have some reliable, you know, some kind of explanations as to why they're holding more inventory.
51:40Amrita Roy:And that needs to match with their product roadmaps. Maybe they're holding more product and they're seeing some general launch that's coming in over the next few months, which is why the inventory levels are moving up. But that sort of needs to flow through in revenues in the coming quarter or two. But, you know, in terms of just the semiconductor company, I would say that I look at these two companies. Outside of that, it's just also gross margins, I think, for software, but it's also ROI. It's also a backlog. So these are big ones.
52:13Uttam Dey:Just to add a little bit more to Uttam's, not necessarily to metrics, I think to the general backdrop of how to think of market. markets. Like I said, I think we are definitely in a higher for longer narrative right now, now that the supply led inflation story has, you know, settled in, and we are going to see inflation tick up in the long term. And until we see sort of a peaking of inflation or markets realizing that inflation is now peaked, and it's about to head lower in any sort of bounce would likely be a tradable one. So during this period of time, I have like two things to say, which is, A, it's like the way we are sort of navigating is, you know, understanding where to raise cash, like cash preservation versus sort of structurally identifying which companies to invest in, given the weakness that we are currently seeing or the mispricings that we are currently seeing.
53:10Uttam Dey:So it's really important, therefore, to manage the risk, the portfolio risk and the risk management in general, between cash preservation and new investments. That's one. And the second thing that I would like to just talk about is that I think, yes, we are hardcore AI investors and we are bullish on the thesis and what it brings over the next many years. But the AI story, I think, is also entering sort of a messy in-between stage right now where, again, like this conversations about ROI on AI CapEx are coming in more and more. In fact, I think this Q1 earnings, there's going to be a lot of focus, a lot of scrutiny being placed on like CapEx, margin pressures and things like that.
54:03Uttam Dey:And usually this is the time where there is often a mismatch between when investors are expecting the revenue to show up versus when revenue actually shows up. Like the timing, there's often a timing mismatch. The revenue eventually shows up if it's an end and we believe it will. But because of the timing mismatch, there can often be like very hard, very hard periods of like sharp, long term drawdowns before the narrative again picks up speed. And I think that's something to keep in mind when building positions and, you know, investing in AI and the tech space in general at the moment. That's all I have to add.
54:42Uttam Dey:Don't invest first and ask questions later. Ask questions first and then invest. Yes. Uttam and Amrita, I'll let you both have the last word. You can follow each of them, both of them on Seeking Alpha, but happy for you to share your substacks as well and where listeners can get in touch with you. So yeah, we have been publishing our research on Seeking Alpha and Substack. We would love for you to follow us on both. We do like a top-down research across the AI supply chain. We are very passionate about the space and finding the winners and also helping you identify losers ahead of time. We are very passionate about this space and will continue to be involved in it for the foreseeable future.
55:25Uttam Dey:And so please follow us. And you're so excited to be part of Seeking Alpha as well for giving us the opportunity to be part of this podcast today. Couldn't be happier and hope you guys enjoyed our conversation. Just a reminder, anything you hear on this podcast should not be considered investment advice. This is for entertainment purposes only, and you should seek advice from a licensed professional before investing.
55:46Amrita Roy:If you enjoyed the episode, leave a rating or review on your favorite podcasting app, and we'll see you soon with a new episode.
From the publisher
Show Notes:
Why Meta Is The Best Positioned Hyperscaler In 2026
Top 5 Stocks For AI's Optical Revolution In 2026
Episode transcripts
For full access to analyst ratings, stock and ETF quant scores, and dividend grades, subscribe to Seeking Alpha Premium at seekingalpha.com/subscriptions

