For most of the last decade, forecasting IPv4 prices meant drawing a line up and to the right. Supply was fixed, demand was growing, and every year the number got bigger. That model worked until it didn't. In 2025 the market fell — hard in places — and anyone whose budget assumed permanent appreciation got the forecast wrong in the expensive direction.
Scarcity is real and permanent. It just turns out scarcity sets a floor, not a trajectory.
What actually happened to prices
A single IPv4 address peaked around $60 during the 2021–2022 squeeze. In 2026 the range is roughly $11 to $32, depending on block size, registry region and the reputation history attached to the space.
The correction was not uniform, and the shape of it is the most useful thing in this whole article:
- Large blocks fell hardest. A /16 now trades near $20 per address.
- Small blocks held. A /24 still trades around $25–$35 per address, which puts a single /24 somewhere near $6,400–$9,000 as a one-time purchase.
That inversion — small blocks holding value better than large ones — looks strange until you remember how the routing table works. Most networks on the public internet will not accept a prefix longer than a /24. Announce a /25 and it is filtered at the edge, which makes the /24 the smallest independently routable unit of IPv4. Below that, the address space stops being useful for the thing most buyers want it for.
So a /24 is not just a smaller /16. It is the atomic unit, and it is the only size a small hosting company, a VPN operator or a single-rack deployment can actually use. Demand at that size has a floor that demand for a /16 does not.
The supply side has not moved, and will not
Every regional registry is out. The dates are worth knowing because they explain why no policy change is going to reopen the tap:
- IANA allocated its final blocks to the registries in February 2011.
- APNIC reached its final /8 in April 2011.
- ARIN exhausted its free pool in September 2015 and has run a waiting list ever since.
- RIPE NCC ran out in November 2019.
- LACNIC and AFRINIC followed in 2020 and 2021.
There is no unallocated pool left to release. Everything that trades now is space someone already holds, which means the entire market is a secondary market and price is set by how willing existing holders are to part with it.
That is the part the old "fixed supply, rising demand" model missed. Supply is fixed in the sense that no new addresses are minted. It is not fixed in the sense that holders can decide to sell — and when prices are high enough, they do. The 2021 peak pulled a great deal of dormant space onto the market, and that supply is what broke the price.
Policy sets the friction, not the price
Registry rules do not determine what a block is worth, but they determine how quickly it can move, and that shapes volatility.
In the ARIN region, transfers happen under section 8 of the Number Resource Policy Manual — 8.2 for mergers and acquisitions, 8.3 for specified recipients inside the region, 8.4 for inter-RIR transfers. Recipients must demonstrate need. RIPE takes the opposite approach: no needs justification for transfers, but a 24-month holding period before a block can be transferred again, which deliberately dampens flipping.
Those two philosophies produce different market behaviour. RIPE's flat, needs-free model made it the most liquid region and a magnet for inter-RIR transfers out of ARIN. ARIN's justification requirement slows transactions and keeps a portion of supply from ever reaching the open market.
One policy detail matters specifically if you are considering leasing rather than buying: in the ARIN region, address space you lease out cannot be used to justify a future allocation request, and leases do not count as efficient utilisation when justifying a transfer. Leasing is permitted — ARIN's policy manual does not prohibit it — but it does not build any registry-side credit.
Lease or buy: the arithmetic
This is where forecasting becomes a decision rather than an observation. Take a /24 at current market:
- Buy: roughly $6,400–$9,000 up front, plus transfer fees and the registry work. You own an asset whose resale value moved 50% against you between 2022 and 2025.
- Lease: roughly $0.45–$0.60 per address per month for a /24, so about $115–$155 per month. Larger blocks lease cheaper per address — a /19 runs nearer $0.35–$0.42.
At $130 a month, a leased /24 costs about $1,560 a year. Against a $7,500 purchase, the break-even sits somewhere near the five-year mark, before you account for the capital being tied up or the resale risk.
Five years is longer than most infrastructure plans, and considerably longer than most companies can predict their own address requirements. That is the honest case for leasing: not that it is cheaper in total, but that it converts a capital bet on a volatile asset into a monthly operating cost you can stop paying.
The honest case for buying is the mirror image. If you know you need the space for a decade, ownership wins on cost and removes the risk of a lessor not renewing.
What to actually forecast on
Given all of the above, "IPv4 prices will keep rising" is not a forecast. It is a slogan, and 2025 falsified it. Some things worth watching instead:
- Block size, separately. The /24 market and the /16 market now move independently. A single blended "price per IP" figure hides the only trend that matters to you.
- Reputation, which is priced. Two identical /24s do not fetch the same money if one appears on major blocklists. Address history is a real component of value and it is checkable before you commit.
- Regional spread. Inter-RIR transfer volume between ARIN and RIPE is the clearest signal of where demand is outpacing local supply.
- Your own timeline. The lease-versus-buy break-even is roughly five years at current rates. If your horizon is shorter, price direction barely matters.
The uncomfortable part
IPv4 pricing is not transparent. Most transactions are private, brokers publish selectively, and the public figures quoted in any article — this one included — are aggregates that may not reflect what a specific block with a specific history will actually fetch. Anyone quoting you a precise 2027 number is guessing.
What can be said with confidence is narrower and more useful. Supply will not increase. The /24 will remain the smallest routable unit as long as operators filter longer prefixes. Reputation will keep affecting price. And the correction of 2025 demonstrated that a market can be permanently scarce and still fall 50%.
Plan for the floor, not the trend. Verify current pricing at the moment you transact rather than from a figure you read months earlier, and check the reputation of any block before money changes hands — those two habits protect budgets better than any forecast.