The following reflects patterns observed in conversations with multiple investment funds assessing IPv4 as an asset class. It is general market commentary, not investment advice or a recommendation to buy, sell, or invest in any asset.

Anyone who has spent time in the IPv4 secondary market has heard some version of the same pitch: supply is fixed, it cannot be inflated, IPv6 isn’t displacing demand any time soon, and address space has historically appreciated, so buying and holding is a reasonably safe bet. Two of those claims are durable. Total IPv4 supply really is capped, and IPv6 adoption, while progressing, is nowhere near erasing demand for legacy space. The appreciation claim is the one that needs updating.

2025 was not a flat year for IPv4 prices. Large blocks—/16s and up—lost roughly half their value as hyperscalers stepped back from acquiring at scale, based on APNIC’s market data. Smaller blocks, in the /24-to-/17 range, held up better but still softened. Prices have leveled off since, with tentative signs of a turn. Whatever the trend going forward, the takeaway for anyone pricing address space right now is straightforward: appreciation is not something to bank on. What matters is what a block can actually earn once it’s in use.

That reframing exposes a structural issue that the standard pitch tends to paper over: acquiring IPv4 and deploying it are two different processes, running on two different clocks, and treating them as one is where most of the bad assumptions creep in.

Two different processes, two different speeds

Acquisition is fundamentally a capital-and-negotiation problem. Given willing sellers and enough capital, a large block can change hands in a matter of weeks. This is the part of the market that behaves like a commodity trade—price discovery, negotiation, close.

Deployment is not that. A block of IPv4 addresses sitting unrouted earns nothing. Value only starts accruing once someone is actually leasing and routing through it, and getting there requires finding a customer, provisioning the space, and building enough routing reputation that the addresses are usable rather than flagged. None of that compresses just because more capital shows up. It’s an operational process with its own throughput limits, not a financial one.

This distinction matters because it’s easy to model IPv4 like a bond—something that starts earning the moment it settles. In practice, newly acquired space behaves more like a piece of unleased real estate: it sits idle until it’s let, and the rate at which it fills up is governed by market absorption, not by the size of the check that bought it.

Why the deployment side moves slowly

A few structural features keep deployment on its own timeline, independent of any single operator’s efficiency.

The buy-versus-lease decision is a pricing question, not a fixed segmentation. Whether an organization buys address space outright or leases it is mostly a function of relative price—at a low enough lease rate, leasing looks rational for almost anyone who needs address space on an ongoing basis rather than as a balance-sheet asset. That describes a lot of latent demand, particularly among operators running distributed, high-churn infrastructure. But matching newly acquired inventory to that demand is still a one-customer-at-a-time process, and pushing the lease rate down to accelerate it directly erodes the yield the whole case is built on.

Leasing volume is concentrated among a small number of large accounts, and landing those accounts runs on quarter-length sales cycles, not week-length ones. Part of that is simply that the leasing market is young. Part of it is governance: a number of registries still treat address leasing as an unsettled question rather than a clearly recognized right—APNIC’s prop-148 is a visible example of this still being worked out in policy fora—which means a meaningful share of each sales cycle goes into explaining that leasing is a legitimate, durable arrangement rather than negotiating price. A surprising amount of that education addresses a basic misconception: that address space registered in one RIR region can’t be routed and used in another. It can, and always could, but the belief persists.

Where a block sits in the registry hierarchy adds a delay that never shows up in a price quote. IPv4 is administered by five Regional Internet Registries, and in APNIC’s and LACNIC’s service regions specifically, national internet registries sit on top of that structure for some economies. That national layer varies enormously in how it treats transfers and leasing, and it’s invisible until you actually try to move the resource. Some NIRs offer no outbound transfer path at all, per APNIC’s own NIR transfer documentation. Some LACNIC-region economies impose multi-year holding periods before a transferred block can move again. By contrast, ARIN and RIPE NCC—with no national layer, settled transfer policy, and deep liquidity, RIPE especially—are the most predictable regions for redeploying acquired space quickly. Both explicitly permit leasing, RIPE unambiguously and ARIN with somewhat more caveats. The practical upshot for anyone doing diligence: a block priced identically to another may not be the same asset, depending on which registry regime it clears through.

Put together, deployment behaves like a lease-up curve—inventory absorbed gradually against a limited number of large-account cycles and against whatever registry posture it happens to fall under—rather than something that switches on the moment capital arrives.

What this means for underwriting and market behavior

If deployment is capacity-gated rather than capital-gated, sizing an acquisition to the opportunity rather than to realistic absorption creates two separate problems.

The first is a cash-flow mismatch: a model that assumes instant lease-up is not a stress-tested model, it’s an aspirational one. Comparing that model against one built on a realistic absorption curve—and being honest about how wide the gap is—tells you a lot about whether the underlying thesis is sound or whether it’s leaning on absorption capacity that doesn’t exist yet.

The second is a market-impact problem, and it’s separate from deployment entirely. IPv4 remains a thin market with a finite number of willing sellers in any given year. A buyer trying to acquire a very large position quickly is a concentrated source of demand hitting limited supply—the textbook setup for a large buyer moving the price of the thing they’re buying. Acquiring quickly doesn’t speed up deployment, and it can raise the entry price for the same inventory that deployment then has to absorb.

The metric that actually matters

The number that determines whether an IPv4 position performs isn’t the acquisition price on its own—it’s the ratio of acquisition price to achievable lease rate. That ratio functions much like a cap rate in real estate: it’s what determines payback and yield from income alone, independent of any appreciation. When acquisition prices fall while lease rates hold—roughly the situation coming out of 2025—that ratio improves, and the economics get better. When acquisition prices climb faster than lease rates, the same capital buys less yield, and the case has to lean more heavily on appreciation, which is a harder argument to make than it was two years ago.

Because deployment speed is ultimately delivered by whoever operates the space, the operator matters as much as the acquisition thesis itself. Worth asking directly: how many of the largest organizations in the target market does the operator already serve, what does its routing reputation and compliance posture look like, and how long does newly acquired space typically take to become fully productive under that operator’s management?

None of this makes IPv4 a weaker asset than the standard pitch suggests—it makes it a different one, with a return profile that has more in common with an infrastructure project carrying a lease-up period than with a fixed-income instrument paying from day one. Anyone underwriting it as the latter is likely to be surprised by the timeline; anyone underwriting it as the former generally isn’t.

This article is part of a bigger opinion piece. Read the full text on the IPXO’s blog.