💰 Cloudy With A Chance Of Excess

THE MAIN EVENT
Cloudy With A Chance Of Excess ☁️

Meta has spent the last two years building AI data centers at a pace that makes your head spin. This week, it dropped a surprise: it wants to rent out some of that capacity.

The company is reportedly building a cloud business, internally dubbed Meta Compute, to sell access to its AI infrastructure to outside developers. The idea appears to come in two forms:

1. Raw GPU access, similar to CoreWeave.

2. Hosted access to Meta’s own AI models, similar to what Amazon, Microsoft, and Google offer.

That matters because Meta is no longer just one of the biggest spenders in the AI buildout. It may become a supplier too.

Meta shares jumped 9% on the report, while the neocloud names got hit hard. CoreWeave fell about 14%, and Nebius dropped about 17%. Despite the market’s strong reaction, there’s more than one way to read this move.

The Bull Case 🐂

The optimistic read is simple: Meta is trying to turn a cost center into a revenue stream.

Meta spent $72 billion on capital expenditures in 2025 and expects $125 billion to $145 billion in 2026 capex, driven primarily by AI infrastructure. That is an enormous bill, even for a company with Meta’s cash flow.

Selling excess compute could help offset that spending while internal demand continues to grow. Zuckerberg has already suggested this was possible, saying cloud compute is “definitely on the table.” He also said companies ask Meta for compute “almost every week.”

That is the key bull argument. Meta isn’t stuck with unused infrastructure — it just built ahead of demand and can monetize the gap.

There is an obvious historical parallel: AWS. Amazon turned internal infrastructure into a cloud business that generated $129 billion in revenue and $46 billion in operating income in 2025. Meta investors would love even a slice of that action.

The Bear Case 🐻

The less flattering interpretation is that Meta may have overbuilt.

If Meta suddenly has capacity to rent out, investors have to ask whether this is savvy monetization or the first sign that the buildout is running ahead of demand.

That concern matters because the AI infrastructure boom has become a major economic engine. The St. Louis Fed estimates that AI-related investment accounted for 39% of U.S. GDP growth in the first three quarters of 2025.

If that spending slows, the pain wouldn’t stop at Meta. It could hit chipmakers, memory suppliers, power companies, data-center builders, and the neoclouds.

The Neocloud Squeeze

The clearest near-term pressure point is the GPU rental companies.

CoreWeave and Nebius are not just competitors in this story. They are also Meta suppliers. Meta has a $21 billion agreement with CoreWeave and a deal worth up to $27 billion with Nebius.

In other words, Meta may be funding the neoclouds today while preparing to compete with them tomorrow.

Wall Street is split on how serious that threat is.

Bernstein called Meta’s cloud push “problematic for CoreWeave,” arguing it is “only a matter of time” before hyperscalers compete directly with GPU-rental firms.

Analysts at Rosenblatt Securities pushed back, calling the selloff a “buying opportunity” because GPU shortages are still “the norm right now across the industry.”

The Bubble Question

The strongest argument against the existence of an AI bubble has been this: unlike the dot-com era, the industry has not been building in search of demand. It has been racing to catch up with demand.

During the dot-com boom, telecom companies laid huge amounts of fiber ahead of actual internet usage. Much of it became “dark fiber,” meaning cable that had been installed but was sitting unused. The internet eventually grew into that capacity, of course. But the timing gap was brutal, and investors paid for it.

AI has looked different so far. The market has been defined by scarcity:

• Not enough GPUs

• Not enough power

• Not enough data-center capacity

• Not enough time

That is why Meta Compute is such an interesting signal.

If Meta is monetizing a temporary capacity gap while its own AI demand ramps, this is disciplined. Data centers don’t arrive in perfectly measured increments, and renting out spare capacity in the meantime is better than letting expensive chips sit idle. But if Meta miscalculated demand by years, not months, the read changes.

For now, Meta Compute is just a plan. The key question is whether Meta can land customers, show real pricing, and prove this is more than a way to soften the optics of a massive AI bill.

One cloud product won’t make or break the AI boom. But if excess compute becomes a broader theme, it could rattle the entire infrastructure trade.

MARKET MOOD
Dow’s Change Lifts Google, Sinks Verizon 🔁

Winners

AeroVironment ($AVAV) - Market Cap: $9.7B (Week-to-Date: +38.4%)

The earnings blowout of the week came from a drone maker sitting right where defense budgets are headed. AeroVironment’s quarterly revenue more than doubled to $642M, while full-year revenue hit $1.98B. And if that was not enough, the U.S. Army awarded AeroVironment a half-billion-dollar contract. Investors see a company moving from niche supplier to defense heavyweight.

Visa & Mastercard ($V / $MA) - Market Cap: $682.3B / $476.6B (Week-to-Date: +7.7% / +8.1%)

Stablecoins were supposed to disrupt the card-network duopoly. Instead, Visa and Mastercard are starting to make crypto look more like a software update and less like a bulldozer. This week, Open USD was announced with 140+ partners, including both card giants, Stripe, BlackRock, and Ripple. If it works, the networks could make payments faster and cheaper… while keeping themselves in the middle.

Alphabet ($GOOG) - Market Cap: $4.36T (Week-to-Date: +6.4%)

Google’s big move was all about supply and demand. Alphabet joined the Dow before Monday’s open, replacing Verizon, and every fund tracking the index had to adjust: buy the new member, sell the old one. That wave of forced buying helped push a company already worth more than $4T even higher. Note to self: index demand can still move very large stocks.

Losers

Teradyne ($TER) – Market Cap: $57.8B (Week-to-Date: -15.5%)

Teradyne got the same kind of index-buying lift we just saw with Google after it joined the Nasdaq-100 in late June. Then reality set in: chip stocks sold off to start July, and Q2 revenue is expected to cool after a blowout Q1. The AI chip-testing business remains strong, but the stock seemed to get ahead of itself. Momentum giveth, and momentum taketh away.

Verizon ($VZ) – Market Cap: $177.7B (Week-to-Date: -8.6%)

This boring ol’ telecom got two tests in one week. First, Verizon was removed from the Dow, creating the forced selling we mentioned above. Then SpaceX reportedly told investors it plans to launch a Starlink-branded mobile service for U.S. consumers, a direct shot at the wireless giants. Verizon still has the network, customers, and cash flow, but investors now have to wonder: how safe is that moat?

Equinix ($EQIX) – Market Cap: $98.8B (Week-to-Date: -8.2%)

Equinix owns data centers, the buildings behind cloud apps and AI workloads, but it is still a capital-heavy REIT, which means interest rates matter. The 10-year Treasury yield moved higher this week, and Meta’s plan to sell excess AI compute made the market feel more crowded. Nothing dramatic broke at Equinix. The stock just got hit by higher-rate math and a less-forgiving AI trade.

CHART OF THE WEEK
Nine Sectors Finish The First Half Higher 🟢

We're officially halfway through 2026, so here's the sector-by-sector scoreboard. A few things jump out.

Tech ran the show, but it had company. Technology led the pack, up about 25% on the back of the AI trade. But this was not a one-sector rally. Energy (+19%), Industrials (+19%), and Materials (+15%) all posted monster first halves too. The old economy quietly kept pace with the robots.

The biggest loser might surprise you. The worst sector of the half was Communication Services, down about 7%. That's the corner of the market that houses Meta, Alphabet, and Netflix.

Breadth was better than expected. 9 of the 11 sectors finished the half in the green, so the gains ran wide. The equal-weight S&P 500 even beat the market-weighted version, meaning the average stock, not just the mega-caps, had a strong six months.

So that's where we stand at the break. The bigger question is what happens next, and that’s where you come in.

6 questions, no wrong answers. Next week, I'll show you exactly where the community landed: bullish, bearish, and everything in between. Don't sit this one out.

Take the 60-second survey here →

FAST FACTS
Bots, Bosses, And BBQ 🌭

🤖 A Bot Army Faked A No. 1 Hit: Spotify scrubbed 500,000+ streams from a song after bettors used bots to rig it to the top and cash in on Kalshi. [Read]

🤦 Bosses Regret Their AI Layoffs: 55% of leaders who swapped workers for AI now admit it was a mistake, and some are already rehiring. [Read]

🍔 Your Barbecue Bill Hit A Record: Feeding 10 at a July 4th cookout now runs $73.82, the priciest since the Farm Bureau started tracking in 2016. [Read]

🏠 Homebuyers Are Back In Charge: The typical U.S. home takes about 18 days to sell now, 3 times longer than the bidding wars of 2022. [Read]

🥇 Gold Wraps A Brutal Quarter: Gold bullion sank about 16% last quarter, its worst since 2013, thanks to a strong dollar and rate-hike fears. [Read]

🎮 PlayStation Goes Disc-Free: Sony announced plans to stop making physical PlayStation game discs in 2028, with digital already making up 85% of sales. [Read]

🏙️ How Much It Takes To Afford Rent: Renting a modest two-bedroom now means earning $34/hr nationally, and north of $49/hr in California and Hawaii. [Read]

WORDS TO REMEMBER
Investing Starts And Ends With Self-Control 🧠