# Google Sold Its Satellites and Kept the Photos
> We pulled Alphabet's $4B stock portfolio: 43% points at space, AST SpaceMobile is a 16x, and a $900M SpaceX stake is now worth ~$100B. The product is data.
- Author: Barebone Research, Barebone AI
- Published: 2026-06-08
- Canonical: https://barebone.ai/resources/google-13f-portfolio-space-data-playbook
- Publisher: Barebone AI (https://barebone.ai)
---
## The Company That Sold Its Satellites and Kept the Photos
In 2014, Google paid **$500 million** for Skybox Imaging, a satellite startup whose pictures helped keep Google Maps fresh. Three years later it sold the whole operation — the team, the technology, and a fleet of seven high-resolution SkySat satellites — to a small imaging company called Planet Labs.
On the surface, a retreat. Read the terms and it was the opposite. Google didn't take cash. It took **a large equity stake in Planet**, and signed a multi-year contract to buy the photos its own former satellites would keep taking.
Sell the hardware. Keep the data. Own a piece of the company you now pay.
Nine years later, that swap sits near the top of a stock portfolio most people don't know Google runs. We used Barebone to pull Alphabet's latest 13F — the quarterly SEC filing where large institutions disclose their US stock holdings — along with every press release and deal document attached to the names inside it. The filing, submitted May 7 with holdings as of March 31, shows **26 stocks worth $4.02 billion**.
One number jumps off the page: **43% of the book points at space** — just two companies, Planet Labs and AST SpaceMobile. Add the SpaceX stake that never appears in any 13F, and Google's space bets are worth roughly **25 times its entire public portfolio**.
What follows is the story of why — including a roughly **$51 million** convertible note that became **$823 million** in twenty-nine months, and a **$30 billion** contract disclosed three days before this article.
## The $4 Billion Tell
Here is the top of the book, with the detail most coverage skips: what came stapled to each stake.
| Holding | Filed value (Mar 31) | % of book | What came attached |
|---|---|---|---|
| **CME Group** (CME) | $1.03B | 25.6% | 10-year deal moving CME's markets onto Google Cloud (2021) |
| **Planet Labs** (PL) | $985M | 24.5% | Multi-year imagery purchases (2017); Suncatcher satellite contract (2025) |
| **AST SpaceMobile** (ASTS) | $741M | 18.5% | Android connectivity collaboration (2024) |
| **Revolution Medicines** (RVMD) | $320M | 8.0% | — |
| **Arm Holdings** (ARM) | $297M | 7.4% | — |
| 21 smaller positions | $643M | 16.0% | Mostly venture-arm leftovers |
Two-thirds of the money sits in three names. The tail — Freshworks, UiPath, Tempus, and a long shelf of clinical-stage biotech — is mostly residue: companies Google's venture arms backed privately that later went public. Those positions arrived by accident of IPO timing.
The top of the book is different. It was built on purpose, and every big position came with a commercial agreement attached. CME Group is the cleanest example, and the newest: Google put **$1 billion** into non-voting convertible preferred stock in November 2021, alongside a ten-year partnership to move CME's trading infrastructure onto Google Cloud. In the March filing that relationship surfaces as **3.48 million common shares worth $1.03 billion** — the main reason the whole portfolio jumped from $2.6 billion to $4 billion in a single quarter.
A fund buys a stock and hopes the company finds customers. Google buys a stock and *is* the customer — or the vendor, or the distribution channel. That's the tell, and once you see it, the space concentration stops looking eccentric.
Because the theme of this book isn't space. It's what space produces.
## Invest. Become the Customer. Take the Data.
Go back to the Planet deal, because the mechanics are the whole playbook in miniature.
When the Terra Bella sale closed in April 2017, Google got equity in Planet and Planet got a guaranteed customer: Google, contractually committed to buying SkySat imagery for years. Every dollar Google spends on those photos becomes Planet's revenue. The revenue supports Planet's valuation. The valuation marks up Google's stake.
One payment, two pockets — and the data lands in Mountain View either way.
Eight years later, Google ran the same play with the same partner, at a higher altitude. In November 2025 it unveiled **Project Suncatcher**, a research moonshot to put TPUs — Google's in-house AI chips — on solar-powered satellites, with a long-run sketch of 81-satellite compute clusters in orbit. The partner building and operating the first two prototype satellites, targeted for launch by early 2027: **Planet Labs**.
So Google now pays Planet twice — once for pictures of Earth, and once to build the orbital test bed for its space data-center ambitions. Planet, for its part, grew revenue **42% year over year** last quarter, to $94.2 million, and raised full-year guidance to **$425–441 million**.
Notice what Google accumulates in each transaction. Not launch capacity, not hardware margins — coverage: a continuous, daily-refreshed photograph of the planet, flowing into the company that turns raw information into products. YouTube was every video humanity would make. Android was a sensor in three billion pockets. A satellite constellation is the same acquisition with a longer lens.
Concentrated customers are usually listed as a risk factor. A concentrated customer who owns your equity and needs your data is something else: a sponsor whose incentive is your survival.
## The Cell Tower in the Sky
The second space position shows the playbook running at full speed, with a timestamp on every step.
On January 18, 2024, AST SpaceMobile — a pre-revenue company building satellites that function as cell towers in orbit, beaming broadband to ordinary smartphones — announced a strategic package of up to **$206.5 million** from AT&T, Google, and Vodafone. The core was **$110 million of subordinated convertible notes, convertible at $5.75 per share** — a 39% premium to the prior close. AT&T attached a $20 million revenue commitment tied to the first five commercial satellites; Vodafone a $25 million minimum. And Google attached the thing only Google could:
> "Google and AST SpaceMobile agreed to collaborate on product development, testing and implementation plans for SpaceMobile network connectivity on Android and related devices."
That's the entire flywheel in one sentence of a press release. Capital, plus a path into the operating system running on roughly three billion devices. Worth being precise about what that is: an agreement to collaborate on *plans* — not, despite how the story gets retold, AST already wired into Android as shipped product. The gap between those two states is where the execution risk lives.
The market didn't wait for the distinction. Alphabet's 13F shows **8,943,486 shares** of ASTS — multiply by the $5.75 conversion price and you get almost exactly $51 million, Google's slice of those notes. At ASTS's June 8 close of **$92.06**, the stake is worth about **$823 million**.
That is **16 times** the conversion price in twenty-nine months. The viral version of this story says Google's AST bet 10x'd. The viral version undersells it.
## The Bet That Isn't in Any Filing
Everything above is the visible portfolio. Google's biggest space position has never appeared in a 13F, because 13Fs only cover public securities — and until the end of this week, this one isn't.
In January 2015, Google invested **$900 million** in SpaceX for an initial stake of about 7.5% — almost a year before a Falcon 9 booster had ever landed. The position then sat in the dark for a decade, slowly diluting, until IPO season forced it into the light: a SpaceX regulatory filing put Alphabet's stake at **6.11% as of the end of 2025**, likely closer to 5% after February's all-stock merger with xAI.
Run the arithmetic against the reported **$1.75 trillion** IPO target: somewhere between roughly **$87 billion and $107 billion**. Call it on the order of $100 billion — about a **hundredfold return**, and roughly 25 times Alphabet's entire public stock portfolio, sitting in a single private line item.
Then last Friday, June 5 — listing days away — SpaceX's amended IPO filing disclosed something that makes the old playbook look quaint: Google will pay SpaceX about **$920 million per month** for AI computing capacity, roughly **$30 billion** from October 2026 through June 2029, renting on the order of 110,000 NVIDIA GPUs at the Colossus complex in Memphis — the data center SpaceX inherited when it absorbed xAI.
Invest in 2015. Become the customer in 2026, eleven years later, three days before the IPO. The same playbook, with three more zeros.
The chart above is the uncomfortable scoreboard: of Google's three contract-paired bets, the best one — by a factor of six — is the one no retail investor could touch, priced not by a market but by a reported IPO target.
Which is exactly where the skepticism should start.
## Where the Flywheel Slips
**First, the circularity.** When the sponsor's spending is the holding's revenue, one dollar gets counted twice. Google pays Planet; Planet books revenue; the stock re-rates; Google's stake marks up; headlines declare Google a brilliant investor. Every step is real and disclosed — and the loop still only spins while the sponsor keeps paying. The $30 billion SpaceX contract has the same property in fine print: either side can terminate after December 31, 2026 with 90 days' notice. The headline is $30 billion. The committed portion is a fraction of that.
**Second, the marks are paper, and the paper moves violently.** Three days before this article, the space book demonstrated it.
On Friday June 5, Planet fell **26% in one session** — directly after reporting that 42%-growth quarter and *raising* guidance. The trigger wasn't the business; it was a freshly filed **$1.5 billion at-the-market equity program** (a standing license to sell new shares into the market) landing on a stock that had run vertically, plus a **$138.9 million GAAP net loss** swollen by warrant revaluations. ASTS fell **12.7%** the same day and closed June 8 at $92.06 — 31% below its late-May closing high. Pick your culprit: dilution, profit-taking, or the optics of Google writing its biggest space check yet to the company whose Starlink direct-to-cell service is AST's most obvious competitor. The tape doesn't label its reasons. Four sessions took Planet down 32%.
Google's blessing, in other words, is in these prices. Cash flows mostly aren't yet — Planet still loses money on a GAAP basis, and AST is still pre-scale, paying interest on those notes in kind. Sponsorship re-rates a stock; it doesn't make the satellites cheaper to build.
**Third, you can't copy the trade that matters.** The hundred-bagger is private until the listing — expected within days — and illiquid even for Google: eleven years without an exit, and standard post-IPO lockups will govern what Alphabet can actually do next. A ~$100 billion mark on a reported target is not money; it's arithmetic.
**And fourth, none of this moves Alphabet.** A $4 billion public book — even a brilliant one — is a rounding error against Alphabet's market value. This portfolio tells you about Google's strategy, not about Google's earnings.
## What This Means
Read a tech giant's 13F the way you'd read its org chart, not its brokerage statement. Nothing in Alphabet's filing is a stock tip. It's a list of capabilities Google decided to rent rather than rebuild — and the signature is always the same: **equity plus a contract**. CME got the check and the cloud migration. Planet got the check and the imagery contract. AST got the check and the Android collaboration. SpaceX got the check and, eleven years later, a $920-million-a-month customer.
The working framework:
**Watch the contracts, not the filings.** The next 13F (due mid-August) will be stale on arrival. The repricing happens when a commercial agreement drops — Suncatcher in November, the compute deal last Friday. Each one tells you which data stream Google wants next.
**Watch whether plans become products.** AST's Android agreement is still "implementation plans." Planet's Suncatcher prototypes are due by early 2027. The gap between announcement and delivery is where these valuations get tested — in both directions.
**Watch the sponsor's own behavior.** Whether Planet leans on that $1.5 billion ATM, whether Google's stake survives SpaceX's lockup expiries intact, whether the next filing shows new contract-paired names. Kingmakers reveal their priorities by where they stop writing checks, too.
The deeper pattern is twenty years old. Google has never really bought companies, or satellites, or GPUs. It buys streams of information at their source — every video, every pocket, every square meter of Earth, every phone outside cell coverage — and it has learned to get paid in equity for becoming the customer. Space is simply the newest faucet.
The question the next filing will answer is the one worth sitting with now: which independent data stream does Google need next — and who's positioned there before the check arrives?
---
*Data: Barebone | Sources: Alphabet Q1 2026 13F-HR (SEC EDGAR, filed May 7, 2026), AST SpaceMobile January 18, 2024 press release (SEC 8-K exhibit), Planet Labs Q1 FY2027 results and 2017 Terra Bella acquisition releases, Google Project Suncatcher announcement (November 2025), SpaceX IPO filing coverage (Bloomberg, Reuters, CNBC, June 2026) | Prices as of June 8, 2026 close*