Listen to the audio version of this article (generated by artificial intelligence).
Earnings season has a way of cutting through the noise. This week, he divided the economy into two parts.
On the one hand: AI infrastructure stocks report impressive results, increase guidance, and struggle to keep up with demand.
On the other hand: traditional companies lower their expectations under the weight of rising costs.
The same economy. Radically different facts.
The AI boom is not just surviving the current macro-disruptions. It speeds through. Everything tied to old economy inputs – energy costs, discretionary consumer spending, interest rate-sensitive balance sheets – is being crushed under the weight of these factors.
The great “AI bifurcation” is beginning to split the economy – and the market – in two. The gap between them is rapidly widening.
AI infrastructure stocks just posted huge gains
Four companies. Four different angles of the AI infrastructure stack. One unambiguous signal.
Vertev: Demand for data centers exceeds capacity
Vertif (VRT) is a core resource within the AI Infrastructure Group. It makes the cooling and power management systems that keep AI data centers running smoothly. On the company’s first-quarter earnings call, CEO Giordano Albertazzi made it clear: Customer urgency has increased, the scale of deployments is greater, and technical complexity is greater than it was six months ago.
The numbers support this. Vertiv reported organic sales growth of 23% in the first quarter. The Americas – where hyperscale accumulation is concentrated – rose 44% organically. Free cash flow more than doubled 147% year-over-year to $653 million. At $1.17, earnings per share (Earnings per share) beat guidance by $0.19 and were up 83% from last year.
The numbers are strong. And guidance is stronger.
Vertiv raised its full-year EPS guidance to $6.35 – up 51% from 2025 – and full-year revenue guidance to $13.75 billion, reflecting 34% growth. Adjusted operating profit is now expected to reach $3.2 billion, an increase of 53%.
Vertiv has no room to grow faster. GE Vernova, which supplies the power that makes Vertiv’s cooling systems necessary, faces a similar problem.
GE Vernova: The force behind the AI boom
CEO Scott Strazyk opened his latest quarterly remarks by noting that the company’s total backlog has increased from $116 billion at the spinoff to $163 billion today. It now expects the backlog to reach $200 billion by 2027, a full year ahead of schedule.
In the first quarter alone, J Vernova (Jeff) Booked $18.3 billion worth of orders – up 71% year over year – at a book-to-invoice ratio of about 2X. This means that for every dollar of product shipped, customers order an additional $2.
Free cash flow in the quarter was $4.8 billion – More than the company produced in all of 2025. Likewise, data center orders for the company’s electricity segment reached $2.4 billion in the first quarter alone – equivalent to an entire year’s volume for 2025, in a single quarter.
Gas turbine pricing is now 10 to 20 percentage points higher on new orders compared to backlogged orders in the fourth quarter of last year. GE Vernova raised its full-year free cash flow guidance to $6.5 billion to $7.5 billion, up from $5.5 billion previously.
The amount of orders for April alone has already exceeded the entire first quarter.
This is not a normal growth rate. It is the product of a huge and growing wave of demand.
Teledyne: The demand for defense and artificial intelligence is converging
The Iranian conflict that is choking United Airlines is, for… Teledyne Technologies (T.D.Y), revenue catalyst.
The company manufactures sensors, imaging systems, drones and defense electronics. In our latest earnings call, Executive Chairman Robert Mehrabian announced record first quarter sales, record earnings per share, and record operating margins – The company’s 10th consecutive quarter of a book-to-bill ratio above 1.0.
Defense orders are on the rise, as the US government actively invests in responding to the Iranian conflict and broader geopolitical pressures. Counter-drone orders reached “tens of millions” in the first quarter alone. The company is shipping the Black Hornet Nano Drone — the smallest autonomous combat drone on the market — to customers around the world. Their total defense business is approaching $2 billion in annual revenue, growing at high organic rates, and the pipeline is growing.
It turns out that geopolitical instability and demand for AI infrastructure are not competing forces. For Teledyne, they are the same order book.
ASM International: Demand for chips maximizes supply
Then there ASM International (name), a Dutch semiconductor equipment manufacturer whose atomic layer deposition (ALD) tools are essential for manufacturing the chips that power AI models.
ASM reported first-quarter revenue of €863 million (~$1.01 billion) – at the high end of guidance – with a record operating margin of 33.1%. Q2 guidance calls for EUR 980 million (about US$1.15 billion), which is close to the quarterly run rate of EUR 1 billion (about US$1.17 billion).
CEO Hisham Massad wasn’t shy: “We’re fully booked for this year.”
The company has confirmed that it will outperform the broader chip manufacturing equipment market in 2026 – a market that will achieve even greater growth. Gartner Estimates now indicate growth of 25%.
Each part – logic, memory, strength – grows simultaneously, a rare preparation in semifinals. the reason? Chipmakers are moving to increasingly complex transistor designs that require more ASM tools in each new generation. Current advanced chips are already in mass production. The next generation begins trial production in the second half of this year – and ASM has already won key equipment there.
The order runway is not measured in quarters. It is measured in chip generations.
The old economy is under energy and cost pressures
While companies across the AI infrastructure spectrum were posting impressive results, last quarter proved to be a slog for companies mired in the “old economy.”
For example, Sunoco cut its full-year earnings guidance, citing higher costs that it could not fully pass on to customers. Likewise, United Airlines revised its 2026 outlook downward, as it faces weak demand signals and higher fuel costs driven largely by higher oil prices.
These disappointing results are not the result of strategic errors. These companies are exposed to incorrect inputs at a time when the Iran war has led to higher energy costs. In a world of razor-thin discretionary margins for consumers, higher costs translate directly into lower profits.
This is the thesis of “Hormuz Losses” playing out in real time.
High oil prices act as a structural tax on energy-intensive, consumer-facing companies such as airlines, packaging, logistics and shipping companies.
For hyperscalers and AI infrastructure builders, this tax simply does not apply. Their contracts are priced on demand, not jet fuel.
Bottom line: AI infrastructure stocks are declining
The divergence we observe between the “old” economy and the “new” economy is not a quarterly anomaly driven by the timing of backlogs or easy year-on-year comparisons.
AI infrastructure requires years of continuous construction; Orders submitted today will not be delivered until 2027 or 2028. The backlog has been closed. Capital expenditure commitments from ultra-large companies are increasing. Demand from sovereign governments, data center developers and enterprise customers alike is accelerating.
On the other hand, the old economy is navigating a world characterized by rising energy costs, weak consumer demand, and persistent uncertainty linked to geopolitical conflict that has permanently shifted oil price floors. Anyone whose business model touches physical goods, fuel or the price-sensitive consumer is feeling real pressure.
We are watching two movies on split screen. The AI has a better plot, a bigger budget, and all the action now.
The structural tailwinds for AI infrastructure—energy, chips, grid, and defense electronics—are not waiting for macroeconomic stability. They double down on it.
Position accordingly.
Today, capital is flowing into infrastructure.
But the biggest gains tend to come in the next phase, when platforms built on that infrastructure take over.
This transformation is already underway.
And there is one company in the middle of this – still a private company, but moving steadily towards what could be AI’s most significant IPO of the decade.
I found a way to move forward before that moment, before it became obvious.
