· 10 min read

Vantage Data Centers Is Exploring a $100 Billion IPO. That Number Tells You Where AI Infrastructure Money Is Actually Going.

Stack Tools

Reuters reported on August 13, 2026 that Vantage Data Centers is exploring an IPO or sale that could value the company at around $100 billion, citing people familiar with the discussions. If it happens anywhere close to that number, it would be the largest data center IPO on record, bigger than the $15 billion take-private of CyrusOne in 2022, which was itself the largest M&A deal the sector had ever seen. Nobody has priced a deal yet. But the number being floated tells you something concrete about where the biggest checks in AI right now are actually being written, and it's not on chips.

What's actually being reported

I went looking for the original wire story instead of taking the secondhand version at face value, because "$100 billion" headlines invite rounding in whichever direction makes the story bigger. The Reuters piece, bylined by Milana Vinn and Echo Wang, is carefully hedged, and it's worth reading it that way rather than as an announcement.

Vantage, backed by private equity firm Silver Lake and infrastructure investor DigitalBridge, is exploring options including an IPO or sale as soon as next year. Two sources told Reuters the company could raise around $10 billion and pursue a stock listing at a valuation of about $100 billion. A third source said Vantage may instead pursue a sale, including a stake sale. Crucially, the company has held preliminary discussions in recent weeks and met informally with financial advisers, but no formal process has been launched, and the sources explicitly cautioned that timing, structure, and size all remain subject to change. Vantage could decide not to do any of this. Silver Lake declined to comment; DigitalBridge and Vantage didn't respond to Reuters' requests for comment.

For context on where Vantage is starting from: the company has raised roughly $11 billion since late 2023, including a $9.2 billion equity round led by DigitalBridge and Silver Lake, and the company runs 35 data center campuses across North America, Europe, Africa, Asia, and Australia. It also recently partnered with Oracle and OpenAI on a Wisconsin campus tied to Stargate, the SoftBank, OpenAI, and Oracle joint venture aiming to build up to $500 billion and 10 gigawatts of AI data center capacity. None of that is a valuation. It's the scale of physical build that makes a $100 billion number plausible enough for bankers to have the conversation.

Vantage isn't the only one

This is the part that makes the story more than a single company's fundraising rumor. Reuters reported in July that Switch, another large data center operator, had hired banks for an IPO that could raise up to $10 billion at a valuation around $80 billion. CyrusOne, owned by KKR and Global Infrastructure Partners since that $15 billion 2022 take-private, is reportedly meeting with banks including Goldman Sachs and Morgan Stanley for a potential IPO as early as 2027, though sources say the company hasn't settled on a raise size (one source pointed to roughly $5 billion). Singapore-based DayOne has confidentially filed for a US IPO targeting around $20 billion, according to Bloomberg. Four operators, four separate processes, all pointed at public markets or a sale within roughly the same 12-to-18-month window.

That's not a coincidence, it's a sector responding to the same signal at the same time: private equity and infrastructure funds that bought into data center operators years ago are looking at where valuations sit today and deciding this is the exit window. Silver Lake and DigitalBridge didn't build Vantage's valuation with model releases or chatbot subscriptions, they built it by financing land, power contracts, and cooling systems fast enough to keep up with hyperscaler and AI lab demand.

The layer under the layer

I made a version of this argument two days ago about CoreWeave and Nebius, when both companies posted blowout Q2 earnings and their stocks jumped double digits in a single session. That post was about GPU rental pricing: neoclouds are pre-selling capacity years out and raising prices on top of rising volume, which is a demand-outrunning-supply story specific to compute rental revenue.

This is a different mechanism entirely. Vantage, Switch, and CyrusOne don't sell GPU hours. They build and operate the buildings, substations, and cooling plants that GPU-renting companies lease space inside. A $100 billion IPO valuation isn't a bet on next quarter's cloud revenue, it's a bet on decades of contracted power and real estate cash flow, financed the way real estate and infrastructure deals get financed: long-duration debt, long-term leases with credit-worthy tenants, and exit multiples set by comparing to other infrastructure assets, not to software companies. When capital markets money moves at this scale into data center operators specifically, rather than into another model lab or another GPU cloud, it's a statement about where investors think the durable, physical bottleneck sits. Land with power access and permitting already sorted is scarcer and slower to reproduce than a GPU order, and pricing that scarcity as a $100 billion company is the market's way of saying so.

What this doesn't mean for you

None of this moves what you pay for compute next quarter. If you're a solo operator renting GPU time from CoreWeave, Nebius, Lambda, or a similar provider, a Vantage IPO doesn't touch your invoice, because Vantage isn't your vendor and an IPO valuation is a claim on future cash flows, not a change in today's supply-demand balance for compute. It's easy to read "$100 billion data center IPO" and "GPU prices staying high" as the same story stacked twice, reinforcing each other into an even bigger, scarier number. They're related but distinct: one is about capital markets pricing physical infrastructure over a 10-to-20-year horizon, the other is about who has pricing power in a rental market this year. Conflating them makes both sound more urgent than either actually is for your monthly bill.

What it should change is your sense of how long this buildout is expected to run. Infrastructure investors don't price 20-year assets on the assumption that AI compute demand cools off in 18 months. If Silver Lake, DigitalBridge, KKR, and GIP are all positioning to cash out multi-billion-dollar data center bets at record valuations right now, that's a signal that sophisticated, patient capital is betting the underlying demand for physical AI infrastructure holds for years, not quarters. That's a slower-moving, higher-confidence signal than any single earnings beat.

The honest take

What I'd actually do if I were budgeting compute costs for the next two years: treat this as confirmation that GPU scarcity pricing isn't a temporary blip caused by one bad quarter of supply, and stop waiting for a "correction" that infrastructure investors clearly aren't pricing in either. Lock in workload flexibility rather than betting on a price drop that isn't showing up in any of the capital flowing into this sector right now.

Where I could be wrong: none of these IPOs has actually happened. Reuters' own sourcing says discussions are preliminary, no formal process has started, and Vantage could walk away entirely. IPO markets can also close fast, valuations can get marked down between a leak and a prospectus, and a wave of four data center listings hitting the market in the same window could just as easily depress each other's pricing through oversupply of the same trade, rather than validate a $100 billion number for all of them. If even one of these deals prices well below what's being floated now, that's a real signal too, just the opposite one: that public investors are less convinced about multi-decade AI infrastructure demand than the private owners currently hoping to cash out.

Author

Lukas

Sources

More Stack articles →

Stay in the Loop

Get new posts delivered to your inbox. No spam, unsubscribe anytime.

Newsletter coming soon. Set PUBLIC_CONVERTKIT_FORM_ID in .env to activate.

Related Posts