Portfolio Manager at Purpose Investments
Member since: Aug '25 · 20 Opinions
You have to look at the entire stack of all similar companies together. You have the hyperscalers, the semiconductors, the supporting infrastructure, and the power component. Across a particular stack, look for dislocations where you have these massive selloffs. A lot more air can come out of the market when it comes to multiples.
You will get opportunities to pick up some of these long-term growers, with very strong fundamentals, as they sell off and the multiples compress.
When you look at the dot-com bubble, today it's a different buildout. Back then, it was funded very much with debt. A lot of the companies were taking on very aggressive debt profiles, with very high interest payments.
This time around, we have the hyperscalers that are extremely cashflow generative. They've enjoyed decades of monopoly-like businesses to give them strong cash balance sheets. So a lot of the growth is getting funded with cash.
On the other side, we're also starting to see a lot of the revenue come up. On the cost structure, companies are also starting to optimize. A lot of operating leverage as companies roll out their solutions. These companies are also eating their own cooking, as they implement a lot of these AI applications internally within their own ecosystems. This also adds to the operating leverage.
Very aggressive M&A strategy. Some of the best capital allocators in the business. Optimizes costs in acquisitions to increase margins, but doesn't necessarily invest in growth. So organic revenue has struggled, and we need to analyze how long this will last. Vertical synergy strategy works well.
Key metric to look for is reacceleration of organic revenue, rather than just M&A revenue.
Liquid cooling systems. Capitalizing on data centre buildout. Sells the picks and shovels to the long-term secular trend (he's a huge fan of this type of strategy). Inning 4-5 of the infrastructure buildout, with capex numbers moving up.
Massive opportunity. Trades at lower multiples because margins are a bit smaller and growth is less than NVDA's. Exposure to revenue from data centre systems is higher than peers. Best pure play in manufacturing components within the data centre space.
Sat out capex on data centres and infrastructure that's depleting other companies' cash balances. Time will tell whether this was a good move or not. The big capex spend may not have been the most efficient use of capital.
Core beliefs are free cashflow and earnings. Consistently buys back shares, which enhances return to shareholders. Apple owns the end consumer. Don't count it out yet.
People are now asking whether AI will eat software. There's a case to be made that if you look at individual point solutions, then they will absolutely be decimated. But you still have power, infrastructure, and platform companies.
Look at LPSN, which automated call centres. Low-hanging fruit that got crushed 98% as soon as ChapGPT came out. Market's still trying to figure out what's going to happen to a company like CRM, which has been a behemoth in the space. But its software isn't great, but is expensive. Co-CEO left to start his own company with an AI-first principle, and that's what other companies need to adopt.
Revenue models need to adapt and adjust to this new normal. Per-seat models have to shift to some sort of consumption model, because AI adoption leads to fewer "seats" to sell to. CRM is trying to do this.
An area of software he really likes is infrastructure, everything that powers the back end behind the scenes.
Software as it exists today should be dead soon, as you can build a lot of it yourself. But you still need the back end and infrastructure to support it. Companies need to reimagine their fundamental DNA and their business models.
Power and energy. #2 performer on the S&P over the last year (after PLTR). US power demand for data centres is going from 5% of the total to 12%. So massive injection of power is needed, and nat gas is the solution until nuclear gets going (which will be a while). In Texas, where many data centres are gravitating towards.
You need to focus on the types of queries that go in. If he wants to learn about uranium and nuclear powering data centres, he'll do a deep dive on ChatGPT. But if he needs new shin pads or a hockey stick, he'll go on Google to find a vendor.
So the search volume is changing in intent. Search queries in Google are becoming much more commercial in intent. YouTube is an absolute beast.