Nitin Gupta

teapawt.net/pequityresearch/status… Timing as expected!

I think the market is making a pretty big mistake in how it's thinking about AI ROI. Market seems to be too focused on whether the frontier labs will make money or not and overlooking what's actually happening at the companies using AI. Some fascinating commentary from $ROP's Q3 call and would recommend everyone to give it a listen. According to the management, 17-18 of Roper's 21 software businesses already have agentic solutions in market with ~75% of Centralreach's bookings and ~25% of Aderant's bookings being AI related. The management also stated that at Powerplan, certain workflows that previously took days or weeks can now be completed in minutes. Neil Hunn also mentioned that the customer interest in these products is the strongest he's seen in his career because they're essentially delivering "a button that does magic things." But this is what what I found particularly interesting for the Hyperscaler CapEx. Roper is monetizing these capabilities through consumption based pricing, like their per script and per load fees model, and AI add ons that can help move customers to higher-priced cloud subscriptions. And management says the gross margins on these agentic tools are roughly in line with their existing software margins. So not only are customers seeing productivity improvements, but Roper is finding ways to monetize those improvements without sacrificing its traditional software margins. Roper also mentioned that they don't need the latest frontier models. They're already using cheaper models and have built systems that allow them to switch between providers, including cheaper frontier models as well as open-source models. This could obviously put pressure on pricing and margins for the frontier labs but is extremely bullish for overall token consumption. What @GavinSBaker has been saying for the last few days. There is also another angle here that I don't think the market is appreciating enough. AI is potentially accelerating the migration of enterprise software to the cloud, where companies like Roper can charge higher subscription prices. That means AI could drive both incremental software revenue and additional demand for cloud infrastructure. Not all of those software dollars will flow to the hyperscalers, but the growing workloads still need to run somewhere. This is where the massive CapEx comes in. If AI adoption continues expanding because businesses can earn attractive returns from deploying it, then today's spending on compute infrastructure will start to make a lot more economic sense. Of course, this doesn't guarantee attractive returns on every dollar of capex. But I find it increasingly difficult to reconcile the argument that AI has no ROI with what we're hearing directly from businesses that are already deploying and monetizing it.

$NLST is up 37% since this. The word seems to be getting out. Don't think anyone mentioned this but the management was in Seoul for the Samsung announcement. It is highly probable this trip included discussions with SK Hynix. SK Hynix's license expired in April, and any negotiating leverage they had is effectively gone after the landmark Samsung deal. Netlist also filed an ITC injunction against $MU today, less than a month before the September 9 hearings, this is clearly a major legal escalation. Do wonder if there was some discovery or terms within the Samsung ITC resolution that provided Netlist with new evidence that potentially strengthens their case against Micron?

Zelnick further added on pre-orders: “The answer is, I don't know, and I'm not going to tell you, but I can say that the level of preorders is unprecedented and astonishing. We're very grateful for that, yet they are so unprecedented that we just don't know how it will translate into sales, which is the question I answered earlier. We genuinely don't know. And we just don't believe in claiming victory before it occurs.” $TTWO

$TTWO This explains the reason for the sandbagged guide. “With regard to pre-orders, our pre-orders are exceptional. No one's ever seen anything like this before at Take-Two Interactive Software Inc. or in the industry. That said, you could absolutely be right. We just don't know. Could demand be pulled forward? Sure. And you know, one of the reasons that we're not changing our guidance is to be clear, you know, we haven't sold one unit yet. You can cancel a preorder. So, you know, we around here like we're sort of allergic to victory laps.”

Was listening to the latest earnings calls for the companies I follow and the validation for the CXL market is becoming prettyclear and will be narrative that should attract attention from investors once the market turns. Management teams for both $MRAM and $NLST (pushing competing memory technologies) explicitly highlighted CXL as the future of data center infrastructure. Everspin said this regarding their timeline and the sheer size of the opportunity: "I would say, first of all, that it's a huge market, and I think it's very well-suited for using MRAM technology. We'll have our first proof of concept here at the end of September. That'll actually, hopefully, successfully demonstrate the use of the CXL protocol using the MRAM technology. With that solution, then we'll actually be able to work with $MXL. We believe that we could actually impact the efficiency by almost 15%-25%." Netlist mentioned how critical CXL will be for managing costs: "Today, a lot of things are moving into the CXL. That will be the future, where more memory products, both NAND and DRAM, it will move to the CXL bus, to the CXL protocol. We think in this market where DRAM costs are sky high, our approach will lower the cost of CXL memory significantly." The ultimate read-through here is extremely positive for $PENG. While others are pointing to solutions that are quarters or years out, @WeAre_Penguin has a commercial CXL solution in the market right now which can capture a lot of this demand.

Surprised $NLST isn't up more on this announcement: "Under the cross-licensing agreement, Netlist will receive an upfront license fee of $200 million after withholding of Korean taxes, or $239 million gross. In addition, Netlist will receive 20 recurring quarterly license fees of up to $27.5 million, also net of tax, or $32.9 million gross, calculated in accordance with a revenue-based formula. Over the five-year term, this represents aggregate net license fees of up to $750 million or gross licensed revenues of up to $897 million. Under the supply agreement, Samsung will supply Netlist with up to $300 million of memory products annually and up to $1.5 billion over the five-year term. This supply includes DRAM and NAND raw materials used to build Netlist proprietary solutions, as well as standard memory and storage products from Samsung for resale to our customer base." They basically get almost their entire market cap in the next 5 years and supply to DRAM and NAND which they can resell + the validation and the investment from Samsung.

$DBO.TO $DBOXF search trends even higher than what they were when F1 released last summer.

Interesting quote from Andy Jassy yesterday on how to read Hyperscaler AI capex. $AMZN $MRVL "Servers and networking equipment operate on a shorter cycle. We typically purchase these a few months before putting them into service, so we have strong visibility into customer demand before we trigger the spend. If the demand isn't there, we won't spend the capital." If they ever guide lower on hardware spend, it’s probably an indication that AI demand is falling off. With $GOOG saying that the picture now is better than a few months ago and $AMZN guiding $200B or higher this indicates that demand is accelerating.

$PENG 🐧

Intrigued by the fact that $TTWO is announcing their results on a Friday, Aug 7th. They haven't reported on a Friday in over a decade. Could mean that they are about to drop Trailer 3 or a Gameplay trailer before the call which might lead to a massive upward revision to bookings guidance?!

$TTWO The release of the game is probably going to be the biggest pop culture event this year that most people don’t understand the gravity of.

$BE $BAM was not kidding when they said they are looking to expand the deal by multiples. Kinda surprised how this was well known for almost 2 months but no one paid attention.