Google Stock Fears a Bubble. The $514 Billion Backlog Says Otherwise.

Google announced this week that it is building a "next-generation AI campus" with Morgan State University, a historically Black university in Baltimore. GOOGL ticked up on the news. For a shareholder — or someone weighing whether to buy in — the right reaction is neither "this is why I own it" nor "so what." A single campus is a rounding error for a company worth about $4 trillion. The deal matters for what it exposes about the engine running underneath it, and that engine is the whole ballgame.

To be precise about the deal first: it is not money flowing into Google's income statement in any material way. Morgan State gets access to high-performance NVIDIA GPU compute through Google Public Sector's research program, plus Google Cloud, Google SecOps and Mandiant security, and enterprise Gemini. Google gets a Center of Excellence on a Baltimore campus and a marketing line about "sovereign AI" — the idea that an institution keeps governance and security control over its own AI while renting the underlying compute. No dollar terms were disclosed. Financially, it is a handshake.

What it is, strategically, is a land grab. Google is positioning the collaboration as a blueprint for digital transformation across HBCUs, and Morgan State is using it to chase a jump from Carnegie R2 to R1 research status. One university won't move a $4 trillion market cap. A long tail of universities, hospitals, agencies, and enterprises that each need heavy, governed AI compute — that is a real market, and it is the same machine that just produced the number that matters.

That number is Google Cloud revenue of $24.8 billion in the second quarter of 2026, up 82% year over year. Read the acceleration, not just the level: the quarter before, Cloud grew 63% to $20 billion. An 82% print beat Wall Street's own aggressive forecast near 64%. This is not a decelerating business catching a tailwind; it is a decelerating rival to the contrary — Google is accelerating while its competitors are not. In the same quarter, Amazon Web Services grew 37% and Microsoft Azure grew 43%. Google's cloud unit used to be the clear third player. On growth, it is now the one setting the pace.

Even better evidence of durability sits in the backlog: $514 billion of contracted — but not yet delivered — cloud and AI commitments, up roughly 390% year over year. A backlog is demand already signed, not demand hoped for. When a company guides to more than double the capital spending it did in 2025, that backlog is the answer to the only question that matters: is there revenue standing behind the buildout, or is this a land grab betting on hopes? $514 billion is the strongest available evidence that the answer is the former.

Now the tension, stated plainly, because it is where the real decision lives. Last quarter, Alphabet raised its 2026 capital spending guidance to $195–205 billion — more than double 2025 — and investors sold the stock about 4% on the news, worried that the AI buildout is an unchecked cost. Shares are down roughly 20% from their May high near $409, even as the demand data keeps improving. That is the market pricing in a bubble: machines built faster than they can be filled, a pile of cash turned into data centers whose economics never arrive.

The data is arguing the other way. Cloud revenue accelerated from 63% to 82% while capacity was being added. Backlog nearly quintupled. When a company grows faster than it is spending — when demand keeps outrunning the machines being built to meet it — that is not the signature of a bubble. It is the signature of an asymmetric setup: the market handed you a discount on the stock because it feared overbuilding, while the order book was growing by roughly 390%.

Here is the honest boundary, because no asymmetric thesis survives without one. Google Cloud's growth is fast, but it is fast from a smaller base than its rivals. Only about 19% of enterprise cloud decision-makers run significant workloads on Google Cloud, against roughly 53% on AWS. Google is converting trials into production more slowly than Amazon or Microsoft. That is the whole risk in one line: Google's cloud business grows at an extraordinary rate from a small share, and the $195–205 billion capex bill only earns its keep if that share keeps climbing. Faster growth from a smaller base can be a feature — or it can be a company filling a narrow lane.

The Morgan State deal sits on the same scale. As AI compute becomes abundant — and Alphabet's capex is making it abundant — the scarce thing is no longer the silicon. It is the governed, secure, institutionally trustworthy way to buy and run the stuff. That is exactly the layer Google is selling Baltimore: not flops, but "sovereign AI" you can put a compliance stamp on. The abundance-scarcity math favors whoever owns the scarce layer, and Google is spending two hundred billion dollars a year to own the abundant one and rent out the scarce one on top of it.

What the campus announcement really gives you is a legible, single-campus sample of a much larger machine: sell managed AI compute to whoever needs it, let the demand show up as backlog, and let the acceleration — 63%, then 82% — be the proof that the spending is finding buyers.

The verdict is not "buy the handshake." It's that the handshake is a distraction, and the numbers underneath it are the trade. Google Cloud is growing faster than its rivals while the market discounts the stock over the cost of growth, and a $514 billion backlog says the demand predates the spending. The open question — the only one that matters — is whether that tiny 19% share of the market converts to something closer to the growth rate Google is booking. Until it does, the acceleration is real and the risk is proportional. The data supports the momentum. Whether it supports the multiple at scale is the bet the market keeps trying to settle on your behalf.