Dear Partners,

For the month of April, Caravel returned +2.29% compared to +7.15% for the benchmark (+10.49% for the S&P 500 & +3.81% for the SPTSX). This brings year-to-date total net return to +9.05% for the fund and +6.95% for the benchmark, respectively.

Our recently increased net exposure to stocks, which we mentioned in last month’s letter, benefitted us in April as markets rebounded sharply from the lows brought on by the Iran war. Both ex-resource (+1.6% gross) and resource (+1.2% gross) stocks performed well, as did our merger arbitrage book (+1.1% gross). Our shorts and hedges were a drag on performance (-1.1% gross). We felt visibility was low and dropping with regards to an off-ramp to hostilities that could restore the conditions which had us feeling optimistic coming into 2026.

Since the market bottomed in late March, we have seen an interesting trend drive the nature of the recovery in stock prices, as measured by popular indices like the S&P 500. To oversimplify, recent performance in the US has largely been driven by stocks that are part of the supply chain for artificial intelligence infrastructure. As we have remarked in the past, these companies have grown so large as to have a dominant impact on the direction of the overall index itself. For example, 9 of the top 10 largest weights in the S&P 500 (‘SPX’) are key participants in the AI arms race. These 9 stocks currently represent over 40% of the value of the SPX. Below is a summary of the performance of the index and these 9 stocks since the market bottomed on March 30.

No, the performance of Micron is not a typo. MU shares bottomed around $320 on 3/30 before beginning a blistering rally to over $800 at their peak last week, though they have since reversed course to $680 at the time of writing. I pitched this stock in a job interview ten years ago. In hindsight, I'm a lot less upset about not having gotten the job than I am about not having taken out a loan to buy some shares (they were $12 each).

Our best guess as to what’s going on in the market is as follows. The war in Iran has created a great deal of uncertainty about, among other things, the path forward for global economic growth. As such, once investors got out from under their desks in late March, they gravitated towards stocks that offered a relative abundance of both (certainty and growth). A robust Q1 earnings season no doubt enhanced the market’s confidence in both factors for the AI complex.

Though it is difficult to say who the ultimate winners of the AI arms race will be (consider Anthropic’s ascent over the last twelve months), as we see it, what is not up for debate is the general trajectory. AI models are getting better, faster. Their use cases are multiplying by the day. Every major corporation and country views being on the right side of this development as a matter of existential importance. We don’t believe their concerns amount to paranoia.

The list above contains enterprises of the highest quality. We are always careful not to confuse quality with value. The market positions these companies have developed are nothing short of remarkable. They will supply, benefit from, or in some cases create the AI enabled products and services that will change the world in the coming years. That is why we believe they are getting so much love from the market lately, because they are great assets.

But remember – there are no bad assets in this business. Only bad prices.

As old school investment managers, we believe free cash flow (‘FCF’) is the holy grail of metrics to assess the value of a stock. To illustrate the point I am trying to make, let’s look at two large cap US companies, Alphabet Inc and Exxon Mobil Corp.

Alphabet (the parent company of Google) might be the greatest business ever created. Historically, its monopoly on internet search traffic allowed it to extract robust revenues and high margins from advertisers and other customers. It was highly scalable and converted much of its revenue to cold hard cash on the bottom line of its financial statements. Its growth trajectory has been a function of the explosive growth in internet usage and data globally over the last 25+ years. Though advertising is economically sensitive, Google commanded high multiples of its free cash flow throughout good times and bad. It has spent most of the past ten years trading at a free cash flow yield between 3% and 6% and an average of about 4%, or 25x its free cash flow.

Exxon Mobil is among the world’s largest producers of oil and natural gas. Given the capital intensity of its business and the volatility of the commodity markets into which it sells its products, its free cash flow yield has exhibited a much broader range in the last decade, getting as high as 15% + in 2022 after Russia invaded Ukraine and turning negative in 2020 after COVID shocked demand and oil prices (briefly) fell below zero for the first time ever. Despite the wide range, XOM’s free cash flow yield has averaged closer to 6% (16.7x FCF) over the past decade.

This all makes sense on a backward-looking basis. GOOGL and XOM are very different companies in different industries that have historically warranted different free cash flow multiples.

But something interesting is happening on a forward-looking basis. 

Exxon has always been in the oil and gas business. Once new wells are drilled and begin producing, their output levels begin to decline. This dynamic requires energy companies to drill new wells every year in order to replace the production they lose from those declines, let alone grow their production over time. This shows up in a company like XOM’s annual capital expenditures (‘CAPEX’), which are always significant. In the past ten years XOM’s CAPEX has amounted to 150% of its free cash flow. It is expected to average around 65% over the next three years as the Iran war has led to a consensus view of ‘higher for longer’ energy prices, which would drive XOM and its peers’ free cash flow higher.

GOOGL has historically been in the software business. Relatively cheap to develop and deploy, highly scalable. Its CAPEX has averaged less than 60% of free cash flow over the past decade.

But Google’s business model has changed. It’s not just a software company anymore, it's a hyperscaler. Look at the analyst consensus for GOOGL’s CAPEX (below in red) over the next three years compared to the last ten. 

These numbers are based on Google’s own guidance. Analysts expect the company’s CAPEX to average over 700% of its free cash flow from 2026-28. Importantly, CAPEX is not considered an operating expenditure (‘OPEX’), and therefore is not deducted from GOOGL’s earnings. This is in part why we believe free cash flow is a superior metric to earnings for evaluating stock prices.

It would appear to us that Google’s business model looks much more like Exxon’s now than it did a decade ago based on this increasing capital intensity. Of course, AI demand is growing in a way that energy demand is not. At the same time, AI models are not commodities. Google’s ability to earn a suitable return on the gargantuan CAPEX it is undertaking depends on its ability to win this incredibly competitive race. GOOGL shares have been one of history’s best investments, and we wouldn’t bet against them. At the same time, if you put a gun to our heads, we would rather buy XOM shares at a 6.3% forward free cash flow yield than GOOGL at 0.3%. Old-fashioned till the last.

We continue to focus our stock-picking efforts on sectors that we believe will either directly benefit from, or at least are at a lower risk of being disintermediated by AI and the other durable trends we see in the market.  Over the years we have learned to not chase returns and to stay in our lane.  Our expertise and focus remain on natural resource developers, aerospace and defense companies, biotechnology, and critical infrastructure/supply chain participants. We also continue to run near maximum allocations to merger arbitrage given the robust opportunity set we continue to see there.

We thank you for your continued support,

Jack and Glen