Switch's $80 Billion IPO: A Private-Equity Exit in an AI Data Center Suit

The headline number tells you nothing about the unit economics underneath.

Bloomberg reported that data center operator Switch confidentially filed for a U.S. IPO. Reuters confirmed July 14: Goldman Sachs and JPMorgan are lead underwriters, the offering could raise up to $10 billion, and the company is targeting a valuation close to $80 billion, including debt, as soon as the fourth quarter of 2026.

That $80 billion figure is the first marketing asset investors should subject to scrutiny. Not because the number is fabricated — because the trajectory from where Switch was last publicly visible tells you everything about whose interests this IPO actually serves.

The $11 billion to $80 billion jump isn't growth. It's a leveraged roll-up on the AI narrative.

In December 2022, DigitalBridge and IFM Investors took Switch private for approximately $11 billion, including repayment of outstanding debt. That was May–December 2022, when data center stocks were still riding pandemic cloud adoption, not generative AI mania. Four years later, the same owners are preparing to float Switch at $80 billion including debt — a more than sevenfold increase in stated value.

The capital moves in between tell the real story. In July 2025, Switch retired all $6.5 billion of bank debt from the take-private transaction. Then, since 2024, Switch has raised approximately $4.2 billion across five asset-backed securities (ABS) securitizations — structured debt backed by contracted rental revenue from its data center portfolio. The company also completed a $2.4 billion CMBS (commercial mortgage-backed securities) financing in 2025. Combined, Switch has closed roughly $6 billion of stabilized asset financings.

The owners borrowed against Switch's own contracted revenue to fund growth and clean up the balance sheet, then priced the IPO at seven times what they paid.

That isn't inherently a fraud. Private equity exists to do this. But it does mean the $80 billion figure is the output of a capital-structure engineering exercise as much as a reflection of Switch's operating performance. The ABS activity is the mechanism: securitize contracted cash flows, use proceeds to fund new development, repeat. It's a proven model in data centers. The question is whether the public market will pay the implied multiple once it sees the leverage underneath the headline valuation.

The non-hyperscale positioning is the structural weak point.

Switch markets itself differently from Digital Realty and Equinix. It is positioned as a leader in non-hyperscale data center ABS — the first AAA-rated tranche in non-hyperscale data center ABS was a Switch issuance, closed in October 2025. Its portfolio of 11 data centers across five U.S. markets serves more than 550 customers, with approximately 84% of trust revenue from investment-grade tenants.

The customer list includes Nvidia, Dell Technologies, and FedEx. That's impressive branding. It's also where the business model runs into a problem that Switch's marketing materials don't address.

The industry is bifurcating. Hyperscale customers — the cloud providers building gigawatt-class campuses — now account for the overwhelming share of data center demand. McKinsey estimates global data center demand will reach 219 gigawatts by 2030, quadrupling 2023 levels. The hyperscalers — Microsoft, Google, Amazon, Meta, Oracle — are collectively spending $260 billion on IT capital in 2025 alone, and most of that is going into data centers. The non-hyperscale segment, where Switch lives, accounts for a shrinking slice of that pie.

Public data center operators are responding accordingly. Equinix, trading at $102.9 billion in market cap with an EV/EBITDA multiple of roughly 27.8x, has pivoted heavily toward hyperscale partnerships. Digital Realty, at $71.7 billion market cap and roughly 30.2x EV/EBITDA, runs a similar playbook. Both are trading at approximately 10.5x trailing revenue — multiples that already price in the assumption that hyperscale-driven AI demand is durable.

Switch's non-hyperscale positioning means smaller tenant checks, shorter lease durations, higher tenant turnover risk, and less pricing power. The 550-customer base sounds diversified until you realize that in data center economics, 550 small tenants are harder to manage than five large ones. Revenue concentration isn't just about single-tenant risk; it's about contract stability, expansion rights, and the ability to build out campuses in advance of demand. Hyperscale customers pre-commit to multi-gigawatt builds. Enterprise tenants lease racks and leave when the lease expires.

Switch is being valued at the peak end of the public comps' range while operating in the lower-growth, lower-margin segment of the business.

The valuation math doesn't survive a per-revenue comparison.

Here's the constraint that the $80 billion headline obscures: Switch's revenue isn't public. The company hasn't filed SEC reports since being taken private in 2022. Any investor who buys into this IPO at the reported range is doing so without knowing what the actual earnings base supports.

That's not a trivial gap. Let's work backwards from the comps. Digital Realty trades at roughly 10.6x trailing revenue. Equinix at roughly 10.5x. If Switch is being valued at $80 billion enterprise value and we apply the same revenue multiple, the implied revenue base would be approximately $7.5 billion. We don't know if that's accurate.

Switch was previously public under ticker SWCH on the NYSE before the 2022 take-private. At the time of the buyout, the share price was around $31 (DigitalBridge paid $34.25 per share, an 11% premium). The company's last public filings would have shown revenue in the $1–1.5 billion range — standard for a portfolio of 20 data centers at that scale. If revenue has grown to $7–8 billion over four years, that would represent a five- to eight-fold increase. For context, that would mean Switch grew faster than Equinix or Digital Realty during the same period, despite operating in the less-dynamic non-hyperscale segment.

The valuation requires Switch to have delivered hyperscale-class growth while running a non-hyperscale business model. Those two things rarely coexist.

The ABS machine is impressive. It's also a warning signal about capital structure risk.

Switch's title as the largest single issuer of data center ABS since 2024 is a bragging point in the offering prospectus. It's a risk factor for public shareholders.

Asset-backed securitization works like this: bundle contracted rental revenue from data center leases, sell tranches of that revenue stream to bond investors, and use the proceeds to fund new construction. The AAA-rated tranches — which Switch achieved for the first time in the non-hyperscale space — are the senior, lowest-risk slice. Below those are mezzanine and equity tranches that absorb losses first if tenants default.

The ABS model depends on stable, contracted revenue. That's the whole point. But it also means Switch has already committed a large portion of its future cash flows to bondholders. When you securitize $4.2 billion of rental revenue across five tranches, the operating company's free cash flow is significantly constrained. New development must be funded from non-securitized cash flows or additional debt — which means the leverage cycle continues rather than terminates.

The $4.2 billion in ABS doesn't strengthen Switch's balance sheet. It monetizes future cash flows and shifts risk downstream to the public equity holders who come in at the IPO.

The timing is textbook private equity. The question is whether the market buys the narrative.

DigitalBridge holds 55.8% of Switch. IFM Investors holds 37.2%. Management holds 7.0%. Australian pension fund Aware Super bought a minority stake from the owners in 2023. At the prior private round, investors discussed a $40 billion valuation. Six months later, the IPO range is roughly $80 billion.

The doubling from $40 billion to $80 billion between the private round and the IPO prospectus is standard private equity behavior. It's also the number that public investors inherit. The owners have approximately four years of compound growth, ABS-fueled expansion, and an AI tailwind to justify the step-up. The public buyer gets the rest of the trajectory — and the leverage structure.

For context, the U.S. IPO market has seen $155.5 billion in proceeds year-to-date through mid-2026, the strongest pace since 2021. The pipeline includes SpaceX, Anthropic, OpenAI, and Csquare. The market is hungry for AI-adjacent listings. Switch is a data center company, not an AI company, but the distinction doesn't always matter on day one.

The cross-currents

• Renewable energy since 2016 — Switch has been 100% renewably powered longer than most data center operators have been planning sustainability programs. That's a genuine differentiator for hyperscaler customers with decarbonization mandates. Directional: positive.
• Non-hyperscale concentration — the shrinking slice of the data center pie, with shorter leases and higher tenant churn. Directional: negative.
• ABS leverage structure — $4.2 billion securitized, plus prior CMBS, means future free cash flow is already committed. Directional: negative for equity holders, neutral for the operating platform.
• $6 billion of stabilized asset financings completed — demonstrates market access and execution capability across macro cycles. Directional: positive.
• No public financials — investors are pricing a $80 billion company without audited revenue, margins, or cash flow data. Directional: the single largest risk factor.

Directionally, the positives support Switch as a well-run, well-capitalized operator with genuine scale. The negatives — especially the missing financials and the non-hyperscale positioning at a hyperscale valuation — suggest the $80 billion number is what the owners can ask for, not what the earnings base commands.

You decide which was marketing fluff and which one was analysis.