Nothing is ever as good or as bad as it seems.
While this may sound like a bit of a bummer when times are good, it’s important to remember this when volatility can swing you the other way. The market likes to overshoot in both directions around long-term cyclical trends, and it's important to keep your head about it.
February and the start of March have brought carnage to the stock market, particularly for high-growth darlings that led the surge over the last two years.
There were plenty of macro headlines around slowing GDP growth, tariff wars, a heated Oval Office exchange, and DOGE’s impact on the labour force. Scott Bessent even came out and said that his focus is a lower 10-year. It looks like the US is going to achieve this by just killing growth completely…
Investing now is like driving through fog. Visibility is low, but with a reliable car, we'll reach our destination.
Now, let’s shift gears away from macro and check how the fundamentals of technology companies are shaping up. Could this steep selloff be an attractive entry point? Or is there more pain ahead?
We have had several recent advancements that have stirred up the tech sector already in 2025. Let’s dig in.
The Most Important Question: Are We Overbuilt?
There were two particular daggers right to the heart of the AI theme.
The first was when TD Cowen said the quiet part out loud by releasing a report that Microsoft is cancelling data centre leases. In the report, they say, “While we have yet to get the level of colour via our channel checks that we would like into why this is occurring, our initial reaction is that this is tied to Microsoft potentially being in an oversupply position.”
The second was a Satya Nadella podcast in which he implied that Microsoft was overbuilt. In this podcast, he explained, “I am thrilled that I’m going to be leasing a lot of capacity in ’27, ’28. Because I look at the builds, and I’m saying, ‘This is fantastic.’ The only thing that’s going to happen with all the compute build is the prices are going to come down.”
The market loves a narrative. It’s just as much a social construct as it is a participation in capitalism. These two events violated the present narrative that we need to continue to aggressively build to harness the power of AI.
Tech stocks have been going gangbusters over the buildout of data centres that will power the AI revolution. Megacap tech is slated to spend over $215B on data centres alone next year, up 45% YoY. The knock-on effects are rippling through the entire tech value chain as anticipation builds. Semiconductors, power producers, cloud companies, and anything to do with buildout have been on fire. With everything increasingly interconnected, each node is being scrutinized for vulnerabilities.
The number one question that people are trying to answer right now is: Did we build too much?
Why We Might Not Be Overbuilt
First, I want to highlight other sections from these very same two events because excerpts can be taken out of context. The TD Cowen report also went on to discuss that the Microsoft demand pullback is being more than made up by Oracle’s new capacity builds (MSFT cutting several 100MW+ contracts, while Oracle builds a net new 1.3 GW) and that aggregate demand is still up notably year over year. What’s important to note here is the mix shift that is occurring.
Microsoft dominated the leaderboard for buildout in an aggressive fashion over the last two years. They were building way more and way faster than everyone else. This was largely due to their direct partnership with OpenAI, whereby Sam Altman is constantly in the headlines saying they don’t have enough access to compute. This led to OpenAI partnering with Oracle and Softbank instead of Microsoft on the $500B Stargate contract.
When you have a startup and an established tenured business working together, growth strategies will never directly line up. Microsoft is a massive public company beholden to shareholders who need to see ROI and responsible growth. OpenAI is setting a mountain of cash on fire on its way to achieving the most powerful technical advancement in history.
By reducing their obligations, Microsoft is derisking their capital outlay while at the same time participating in upside from their investment in OpenAI since they get a share of their profits in the future.
Additionally, in Satya’s podcast, he also went on to say, “that infrastructure need for the world is just going to be exponentially growing.” We are now at the beginning of the next wave: inference time scaling. Inference is when the model is "called," and the consumption turns to usage instead of training. Inference time scaling is when you give the model extra time to "think" through its answer recursively, and the quality of the answer drastically improves.
These new reasoning models can easily use more than 100 times as much computing resources as conventional large language models. Inference time scaling = much more compute demand coming = we need to build more.
Another potential inflection point that could lead to an exponential increase in compute demand is agent-to-agent interactions. Right now, we're still figuring out how to implement AI systems into very human processes. It’s not hard to imagine these agents gaining much more autonomy and interacting without prompting from humans. These agents will then transact with each other millions of times behind the scenes to optimize interaction.
An example of this is my personal agent connecting with an airline agent to book a flight, a hotel agent to book a place to stay, and a restaurant agent to book a place to meet for coffee. It will do all this by cross-referencing my calendar, contacts, and email to book meetings with colleagues I might want to catch up with while visiting another city. Seamless, integrated, and proactive engagement all being done in real time.
We’ve Seen This Before…
A lot was said of the Jevons Paradox during the DeepSeek scare. This is the concept that increased efficiency of a product leads to a net increase in total demand. This has been thrown around a lot as of late in the press, but I wanted to give a real-life example. I searched for many use cases and found that Mark Lipacis, a semiconductor analyst at Evercore, did the best job explaining a corollary.
In the early 2000s, virtualization software had recently been introduced by companies like VMWare and Red Hat and was estimated to improve server CPU utilization to >80% from 5% (i.e., by 20x-30x, similar to cost-improvement claims made by DeepSeek). At the time, many investors concluded that the dominant supplier of server CPUs, Intel, would not see any server CPU growth as the installed base improved utilization by 20x-30x.
However, this chart shows that server CPU unit sales continued to increase at a 6%-7% rate, even as server workload virtualization increased to 75% in 2017 from 20% in 2005, supporting the thesis that lower cost computing cycles drive increased demand for those computing cycles.
By The Headlines
What I also find particularly interesting is that the whispers around the concept of overbuilding are beginning to get louder at the same time we are getting headlines like these:
- Meta Plans Record $65bn AI Investment and 2GW Data Centre (TechMag – Jan 30)
- Taiwanese chipmaker TSMC announces new $100B investment in US (Politico – Mar 3)
- OpenAI, Oracle, and SoftBank launch massive $500B Stargate project to expand AI data centre capacity nationwide (APnews – Jan 22)
- India Energy Group Plans 'World's Largest' Data Centre (PowerMag – Jan 27)
- Utilities ramp up $2B natural gas investments to meet surging power needs for new AI data centres (Business Insider – Feb 13)
Wrapping Up…
While technology investors are no strangers to volatility, there is still a visceral response every time you live through it. This is one of those periods. But let’s look at the facts.
We have no shortage of companies expressing medium-term demand for continued buildout. Earnings have reinforced these views as CapEx continues to march higher. NVIDIA's Blackwell chips have been sold out for over 12 months. Most importantly, we are only getting started on AI utilization and implementation.
There will be fits and starts along the way. Macro is still firmly in the driver’s seat, and there could be sustained pressure as we have not quite had a blowout bottoming yet.
However, as Peter Lynch famously says, “Stock prices have a 100% correlation with earnings.” Positioning yourself around companies that drive long-term secular growth puts you in the best position to participate in this earnings growth.
Strong convictions. Loosely held.
– Nick Mersch, CFA
Receive the latest tech insights in your inbox every month.
The content of this document is for informational purposes only and is not being provided in the context of an offering of any securities described herein, nor is it a recommendation or solicitation to buy, hold or sell any security. The information is not investment advice, nor is it tailored to the needs or circumstances of any investor. Information contained in this document is not, and under no circumstances is it to be construed as, an offering memorandum, prospectus, advertisement or public offering of securities. No securities commission or similar regulatory authority has reviewed this document, and any representation to the contrary is an offence. Information contained in this document is believed to be accurate and reliable; however, we cannot guarantee that it is complete or current at all times. The information provided is subject to change without notice.
Commissions, trailing commissions, management fees and expenses all may be associated with investment funds. Please read the prospectus before investing. If the securities are purchased or sold on a stock exchange, you may pay more or receive less than the current net asset value. Investment funds are not guaranteed, their values change frequently, and past performance may not be repeated. Certain statements in this document are forward-looking.
Forward-looking statements (“FLS”) are statements that are predictive in nature, depend on or refer to future events or conditions, or that include words such as “may,” “will,” “should,” “could,” “expect,” “anticipate,” intend,” “plan,” “believe,” “estimate” or other similar expressions. Statements that look forward in time or include anything other than historical information are subject to risks and uncertainties, and actual results, actions or events could differ materially from those set forth in the FLS. FLS are not guarantees of future performance and are, by their nature, based on numerous assumptions. Although the FLS contained in this document are based upon what Purpose Investments and the portfolio manager believe to be reasonable assumptions, Purpose Investments and the portfolio manager cannot assure that actual results will be consistent with these FLS. The reader is cautioned to consider the FLS carefully and not to place undue reliance on the FLS. Unless required by applicable law, it is not undertaken, and specifically disclaimed, that there is any intention or obligation to update or revise FLS, whether as a result of new information, future events or otherwise.