Friday, September 18, 2026

@MaiusPartners Sep 17

Maius on X

3/X The $GNRC Generac / $AMZN AWS Trojan Horse: Tier-1 OEMs ($CAT, $CMI, MTU) are locked in 24 month+ lead-times. $GNRC stepped in with an $8B Amazon backup deal, but GNRC lacks large-bore diesel blocks and enclosure capacity. Who bridges the gap? The Weichai Ecosystem. PSIX monetizes the buildout whether AWS buys through Cat dealers or Generac.

@MaiusPartners Sep 17

Maius on X

4/X The Prime Power Endgame: Backup power is just the wedge. The prize is prime power. SemiAnalysis recently flagged $META and Anthropic deploying 300–500MW sites using 21.9L gas engines— the exact spec $PSIX supplies. As Weichai pushes its larger 12M55 and 20M55 gas engines through North American UL certification, $PSIX becomes the domestic vessel for low-cost, off-grid power generation.

@majgeoinvesting Sep 17

Maj Soueidan on X

I can see that being true. But when you’re playing in the smaller cap universe, the information disconnect edge is crazy huge. Because less investors are looking at these stocks, no one‘s really finding the information. But yes, when people discover the info, the interpretation phase kicks in. With big caps it’s more about the interpretation phase because everyone is looking at the same information. With microcaps, you have two MultiBagger forces happening: Early information discovery + interpretation. By the way, let’s not forget that there’s so much information out there to interpret, that it creates “brain” overload. It’s going to be interesting to see how AI plays a part in solving some of that problem. 👊But generally, I do like your interpretation angle. It’s funny because when you find information arbitrage in microcaps… that’s so obvious… sometimes you have no movement in the stock because people are either interpreting it wrongly or just don’t believe it, etc. By the way, $HPE would be another good Bigcap Case Study you could use to support your interpretation argument. It just sat there at 15 bucks, as people debated their relevance in AI… And then overnight it goes to $50. I rarely buy big caps, but I got lucky and actually bought some call options ahead of the move.

@ACapitalLP Sep 17

A Capital on X

This is actually a great point. $QXO is an opportunistic acquisition vehicle, not a plain vanilla building products company like $BLDR. If it’s just trading with rates, that could be the opportunity… https://t.co/yhGuyFWQAT

@BrokenMoats Sep 17

Broken Moats on X

A few take aways from the $CRM Dreamforce keynote today.... It is difficult to watch at times with all of Marc's promotional showmanship but still worth it to understand how a bunch of legacy software companies are attempting to reposition themselves in terms of the AI era. The sell side will defend this tomorrow, thats the business they are in, but you should look at it for yourself unbiased. Ask yourself this first: .....you can claim SaaSpocalpyse over, the V shape rebound came, ARRs didn't plummet overnight.... That was never going to happen, the newspapers are still here, but their subscriber bases are decimated, or they moved online and the pricing models were drastically decreased. Reach potential increased, but so did competition (exponentially) and barriers to entry were and are radically lower then pre internet. If this was not existential and just a panic and the rebound is entirely justified why did the largest SaaS application company just remove their ui, conceded their metadatabase to their biggest competitor and is undergoing a pricing model change in real time, so change is upon us, time to look what happens next CRM going full force into being basically a database company now, hoping the underling data you built with salesforce over the years and the legacy connection will be enough to keep you around and paying them to essentially host your data for you with a Claude or other ai intelligence on top doing the work. Slack is the most relevant product they have and the biggest sticking point in those conversations, but how long will that last? Zoom should have died off by now, with Teams and others giving away video conferencing which shows you entrenched behaviors and a narrow network effect can defeat economic interests. Time will tell on slack. Ultimately I think a layer like snowflake that can tap data structured and unstructured across applications is the data layer and governance apparatus to build wrappers and dashboards from using multiple Ai models. Versus keeping the data siloed into the old applications from the prior generation and having to access them, or pay them all individually as databases and then building a mass MCP through all them to get to the unified data solution (much easier then ever before to migrate and move data the friction has dramatically been lowered). Seems a lot easier to have one data layer across an enterprise versus taking a salesforce, connecting it to Claude then having to connect that to Gong, granola, gmail, etc for your sales force and cs reps. Then creating channels with workday or paycom and connecting special channels for intuit and adp separately. And so on with each roles/departments specialized applications they use in a given day and giving database, governance, ai training, security etc and having to manage and permission all of these ai/database modules versus a central one stop shop with Databricks or snowflake (id say amazon or google, but no one is going to go all in on one cloud storage for outages and pricing reasons if large enough but I think they have a better argument/permission to win this space if they wanted to then Salesforce or worse a Paycom)....see the snowflake Q2 call to better understand their vision as competing to Marcs strategy laid out today. End of the day, I don't think when you look out three years these SaaS applications have the same relevancy in terms of users but more importantly on pricing (dropping the UI works for AI and looks cool in a demo, but it also erodes your brand and shared user behavior across organizations which was a piece of the stickiness of the solution over these SaaS years). CRM is likely better suited then most of the other applications ironically due to their largley paned at the time acquisition of Slack. I don't think it provides immunity but a solution to cling to in bundled conversations where their numbers may slow slightly less fast then some of their contemporaries. Point solutions continue to be first in line, communications too. These businesses are not impaired to zero today, but their pricing and expansion (new logo and NRR) that is going to be continued to be eroded. Newspapers are still here, magazines too but their subscription based was drastically hollowed out. My view is the SaaSapocalypse was directionally correct and narrative largely holds, the markets in short term may have over run with that theme (some single digit true ebitda multiples) but thats on old business models and we have largely filled the entire gaps to where its attractive to take the outlook of where these businesss are likely to head in the coming years again. They will show amped AI revenues (as they rename and push anything named AI for the sake of all other revenue and rename, bundle to get to say "AI revenues up 80%"....but total ARR, NRR still continue to be flat to declining), they are pushing duration and discounts to achieve this as they themselves scramble with ai models and solutions and rush to make with large over promises, and leave clients with messy solutions that frequently break, have real security issues, and underwhelm the capabilities the sales team pitched (I listen to this daily from my wife whose a cs manager at a software company, and the majority of my friends who work in software in various capacities).

@BlueDuckCap Sep 16 2 clicks

BDC on X

Maybe this earlier post is hyperbolic. But Specs potential is real and the applications are very real and the implied value to $snap is negative. Btw go back and look at all my snap tweets - been as big a bear as any of you. But the worm is turning. Sub business 85% yy $1b+ rr and ads stable and 16% HC reduction to cut $500mn opex and SBC offset by buyback. So there’s a lot to like here and if specs can get a minority investor which i think is much more likely after today then game on. Oh and by the way snap could still cut opex by 50% just to get in line with $rddt on cost/employee.