Two of the most important companies in the AI buildout report today (Wednesday).

Alphabet ($GOOGL) reports after the close. GE Vernova ($GEV) reports before the open.

We published our House Views for members before the prints.

One is a staked call. The other is a deliberate non-call.

That difference is intentional.

Tessara doesn’t force predictions. We stake calls only when we believe the evidence gives us an edge. When it doesn’t, we say so. And publish the framework we’ll use to interpret the result instead.

So here’s how we’re thinking about both prints.

Everyone is probably going to watch one number today: does Alphabet raise its 2026 capex guide?

But we think that’s the least important question in the print.

Whether management leaves the current $180–190B range unchanged or raises it modestly matters far less than what they say the spending is buying.

A dollar flowing into servers, GPUs and deployment has very different implications for the AI supply chain than a dollar flowing into land, power and construction.

One pulls semiconductor demand forward.

The other extends the runway.

An unchanged guide paired with stronger deployment language and confidence around 2027 is more bullish than a capex raise driven primarily by buildings.

We think there are three important questions today

1. Does Alphabet change the guide?

Our model assigns roughly a 40% probability that Alphabet raises its midpoint to around $195B.

Our base case remains reaffirmation of $180–190B.

That is the easiest question to grade, and the least economically important.

2. Is Alphabet executing a larger multiyear AI build than the 2026 guide alone implies?

High confidence: yes.

The evidence already points beyond today’s capex range: an $84.75B financing, a Cloud backlog approaching $460B, strong Cloud growth, and management’s own guidance that infrastructure spending rises again in 2027.

That thesis barely depends on today’s headline number.

3. Which suppliers actually receive the spending?

This is the investing question.

Every dollar of Alphabet capex eventually flows somewhere: accelerators, memory, advanced packaging, networking, power equipment, campuses or construction.

Those paths have very different implications for the companies exposed to them.

Reported capex is only one piece of evidence. We also watch the physical system underneath it.

Our evidence chain runs in this order:

By the time the build reaches reported capex, much of the supply chain has already moved.

That’s why we don’t stop at the guidance number.

Our staked view, published before today’s print, is that the AI buildout remains expansionary even if Alphabet leaves its 2026 capex guide unchanged.

We’ll grade that view publicly after the earnings call, hit or miss.

Our full House View on Alphabet maps each possible outcome to the companies that benefit or get squeezed, from memory and compute through power and infrastructure, together with the evidence that would confirm or falsify each path.

GE Vernova reports before the open.

We are not staking a call.

We looked for one and found no meaningful edge over management’s own guidance. The company has already guided 10–15 GW of new gas equipment contracts this quarter, and our own model lands around 11 GW, comfortably inside that range.

The order number almost everyone will react to is a weaker signal than it appears.

After roughly 45 GW of gas orders over the past two quarters, available delivery slots—not customer demand—appear to be the binding constraint.

Management has already said only about 10 GW of remaining 2029–2030 delivery capacity was still available for sale entering the quarter.

That means an order number inside guidance mostly tells you how many future manufacturing slots GEV chose to release.

It does not, by itself, tell you whether demand accelerated or weakened.

Here’s the part we think the market is underappreciating.

Management expects to finish 2026 with at least 110 GW of contracted capacity.

Today that figure is roughly 100 GW.

The company also expects to deliver about 14 GW over the rest of the year.

That means:

Here’s the insight.

Management has already said only about 10 GW of remaining 2029–2030 delivery capacity is still available for sale. Even including additional fast-turn units, our estimate is that only 13–19 GW of near-term capacity remains.

That leaves several gigawatts of required orders with nowhere obvious to go.

Our inference is that meaningful 2031 delivery bookings likely begin this year.

Management has never framed it that way so far, so this is our insight. If that inference proves right, 2031 (not the quarterly order number) is the real story.

Three things matter more than the headline order number:

Our full GEV House View walks through the reasoning in detail, explains why we logged a forecast of approximately 113 GW year-end contracted capacity, and shows exactly what would change our mind.

Inside Tessara Pro they stay alive.

Members get:

We won’t measure ourselves by how confidently we write. We measure ourselves by whether our work stands up after the outcome is known.

Explore Tessara Pro →

Cheers,

Teng

This article is for informational and research purposes only. It is not financial advice, investment advice, or a recommendation to buy or sell any security. Tessara Research does not publish price targets. The views expressed here reflect our analysis at the time of publication and may change as new evidence arrives. Readers should do their own research and consult a qualified financial adviser before making investment decisions.