From scarce to unwanted
I have been around cellular M2M long enough to remember the first time a supplier sat me down and told me IPv4 addresses were running out. It was well over ten years ago, at one of the big UK M2M aggregators, and the message was simple: public static IPs on SIMs would not be cheap or plentiful forever, so design around them. Most of us nodded, carried on ordering public IP SIMs for remote routers, and nothing much changed for years.
The warning was right, but I now think it was the wrong reason. The public IP SIM is in trouble less because the world has run out of IPv4 and more because a router on a public address can be scanned soon after it connects, probed constantly, sometimes billed for the privilege, and increasingly frowned on by every cyber security agency that has looked at the problem. Scarcity made public IPv4 expensive. Security is now demolishing the case for making it the default.
That raises a question I have not seen anyone ask. If installers and fleet owners move their routers off public IP SIMs and onto private addressing with outbound remote access, those addresses go back to the suppliers that hold them. Could fifteen years of IoT shortage end in something closer to a surplus? This piece follows that story from start to finish, and finishes with what I would actually do with a fleet today.
Public IP SIM vs private IP SIM at a glance
A public IP SIM gives a router or modem an internet-routable IP address, either fixed or changing. Unlike a typical SIM behind carrier-grade NAT, it can accept connections started from the internet, subject to the operator's filtering and the device's own firewall. For most IoT deployments, a private IP SIM with outbound VPN or cloud remote access achieves the same day-to-day result with far less exposure. If you are new to the subject, start with my practical explainer, what is a public IP SIM card, then come back here for the bigger picture.
| Public IP SIM | Private IP SIM with remote access | |
|---|---|---|
| Address | Publicly routable IPv4 | Private or shared, behind NAT |
| Inbound from the internet | Possible, controlled by filters and firewall | Not normally possible directly; depends on provider routing |
| Remote management | Direct, or through a VPN to the router | Outbound VPN, private APN or vendor cloud |
| Uses scarce IPv4 | Yes, one address per SIM | No, shares the operator's public egress |
| Security effort | High: strict inbound controls, patching, monitoring | Lower direct exposure, but devices still need hardening |
| Typical use | Specific inbound or legacy requirements | Most managed IoT fleets |
Contents
- Public IP SIM vs private IP SIM at a glance
- Fifteen years of squeeze: the paper trail
- Why the internet ran out
- Mobile was built on NAT from day one
- What a public IP SIM costs today
- The twist: a public IP is now a liability
- From shortage to surplus?
- Seven alternatives to public IP SIMs
- A worked example: 100 routers
- Is my SIM behind CGNAT? A two-minute check
- What I would do now
- Frequently asked questions
Fifteen years of squeeze: the paper trail
It was not just talk in meeting rooms. Going back through the trade press, regulator documents and supplier blogs, the warnings about public IPv4 in UK M2M are there in print, and they started earlier than I remembered.
April 2011. Stream launched a service called Dina, pitched in M2M Now as a more secure alternative to fixed public IP. Its founding director Nigel Chadwick argued that fixed public IP lacked the security and resilience modern M2M needed. Dina only let a device be reached from an address the customer had linked to that SIM, and blocked everything else. Fifteen years on, that is still the right instinct.
July 2014. Ofcom published a call for input on the Internet of Things. It noted the limited number of unique IPv4 addresses and warned there may be pressure on the pool if IoT grew as expected. It said NAT, the usual workaround, was unlikely to be a sustainable solution at IoT scale, and that moving to IPv6 was necessary where devices needed global connectivity.
August 2014. The squeeze showed up in a UK 4G product. Relish, a fixed 4G wireless broadband provider in London, told ISPreview that global IPv4 exhaustion had forced ISPs onto carrier-grade NAT. It planned to release public IPv4 addresses from a limited block, subject to qualifying criteria, at an extra £10 a month on a £20 service. It was not an M2M provider, but the pattern is exactly what IoT buyers face now: a limited pool, eligibility rules and a premium.
November 2017. Anton Le Saux, head of M2M at Zest4, wrote a piece for customers on public IP versus VPN. He said that worldwide we were at the point of exhaustion
with IPv4 public IPs, and predicted they would probably be gone by mid-2018. He quoted public IPs at £25 or more per month each, against £25 to £60 a month for a VPN. The mid-2018 prediction was wrong; public IPv4 SIMs are still sold today. But his advice, stop asking for a public IP by default, has aged well.
September 2018. Wireless Logic's white paper on SIM selection pushed private static IPs as the IoT pattern, describing a private static address as off the public internet, so the deployed estate behaves like an extension of the customer's own WAN. It also warned that operators can change their internal and CG-NAT addressing without notice, leaving customers chasing new public address blocks through their firewalls.
Read together, these show the industry steering away from public IP SIMs for more than a decade, first on security and reliability grounds and then on cost. Scarcity was the background noise. Security was always part of the argument.
Why the internet ran out
The short version. IPv4 was specified in RFC 791 in September 1981 with 32-bit addresses, giving about 4.3 billion possible values, many of them reserved. Wasteful early allocation and the growth of the internet used the pool up far faster than anyone expected. CIDR in 1993 and private address ranges in 1996 slowed things down. They did not create any new addresses.
Address space flows from IANA to five Regional Internet Registries. For the UK, that registry is the RIPE NCC in Amsterdam. These are the dates that matter.
| Date | Milestone | Why it matters |
|---|---|---|
| Sep 1981 | RFC 791 defines IPv4 with 32-bit addresses | Fixes the ceiling at about 4.3 billion values |
| Feb 1996 | RFC 1918 reserves private ranges | Makes NAT behind one public address normal |
| Dec 1998 | IPv6 specified in RFC 2460 | The successor exists, but is not compatible with IPv4 |
| 3 Feb 2011 | IANA allocates its final five /8s | The global free pool is gone. RIPE NCC receives 185/8 |
| Apr 2012 | RFC 6598 reserves 100.64.0.0/10 for carrier-grade NAT | Operators formalise address sharing at scale |
| 14 Sep 2012 | RIPE NCC starts allocating from its last /8 | One final /22 per member, and only if it already holds IPv6 |
| 25 Nov 2019 | RIPE NCC makes its final /22 allocation | Its pool of never-allocated IPv4 is exhausted. A waiting list opens |
| Feb 2024 | AWS starts charging for every public IPv4 address | IPv4 becomes a visible cost line in the cloud |
| 28 Mar 2026 | IPv6 passes 50% of traffic to Google | Half the measured internet is on IPv6 |
Since 2019, recovered addresses in the RIPE region go out through a waiting list, one /24 of 256 addresses per new member, ever. RIPE's October 2026 member update put 806 members in the queue, with the one at the front waiting 487 days, about 16 months. Everything else comes from the transfer market. Broker data puts small blocks at around $22 to $25 per address in mid-2026, down from a 2021 to 2022 peak when some trades reached around $60, but still enough to make a /24 cost roughly $6,000.
Mobile was built on NAT from day one
Mobile networks have rarely given every SIM a public address. Most use carrier-grade NAT (CGNAT): each SIM gets a private address, from RFC 1918 space or the 100.64.0.0/10 shared range, and the operator translates thousands of subscribers onto a small pool of public addresses at the edge of its network.
What surprises people is how early this started. At the RIPE NCC Days in Sofia in June 2023, the Bulgarian operator Vivacom showed a timeline with its mobile network running CGNAT from the start, in 2007, four years before IANA handed out its last blocks. Among its reasons it listed the increasing cost of buying and renting IPv4. That is one operator, not proof that every network did the same, but it matches what I have seen in the UK: on most mobile services the public IP SIM was the exception, not the rule.
For a phone, CGNAT is invisible. For IoT it has one big consequence: nothing on the internet can start a connection to your device. That is why port forwarding on a standard SIM never works, and why so many installers have lost an afternoon wondering why their CCTV recorder is unreachable. It is also why the public IP SIM became the default answer to "how do I reach my kit". Ask for a fixed public IPv4 address, point your software at it, done.
What a public IP SIM costs today
Public IPv4 on a SIM is still for sale in 2026, and it is not always expensive. What has changed is that it rarely comes on its own terms.
Daisy's Anywhere IoT tariff sheet, correct as of 14 March 2024, lists a public static IP add-on at £7.50 against £2.00 for a private static IP. The public address costs nearly four times as much as the private one, and more than the £5.50 that same sheet charges for a 1GB data allowance. When the address costs more than the data, suppliers will ask whether you really need it.
Things Mobile lists a business public fixed IP at a €25 activation fee and €2 a month, with prices valid from May 2020. The interesting part is the condition: you must declare the public source address of the server that will talk to the SIM, and each extra source address costs €5. Even a cheap public IP comes with inbound access locked to known addresses, the same idea Stream had in 2011.
Simbase charges a daily fee for a public IP that applies whatever state the SIM is in, so a disabled SIM keeps costing money until the address is released. Soracom goes further: its support pages say it is not possible to assign a global IP address directly to one of its IoT SIMs at all. It offers fixed outbound addresses through a gateway and on-demand remote access instead.
Put those side by side and the direction is clear. Suppliers are not refusing public IPv4, but they are pricing it, fencing it and in some cases designing it out. In my experience the questions now asked before a public IP goes onto a SIM are sensible ones: is it genuinely needed, is it in use, and would a private APN, an outbound VPN or a cloud management platform do the same job?
The twist: a public IP is now a liability
This is where the story turns. Even if public IPv4 were free and plentiful, I would not want most routers on it in 2026.
Expect to be scanned
The whole IPv4 address space is small enough to be scanned continuously, and services such as Shodan and Censys do exactly that. A router with a publicly routable IPv4 address may attract automated scanning soon after it connects. Whether scanners can identify its services, and whether those services appear in Shodan or Censys, depends on operator filtering, firewall rules and what is actually exposed. A public address creates the possibility of direct inbound access; the real attack surface is whatever the firewall lets through. If a web interface, SSH or API faces the internet with a weak password or an unpatched flaw, expect it to be tried. In 2025 researchers at Sekoia found more than 18,000 Milesight industrial cellular routers reachable on the public internet worldwide. Of 6,643 they checked, 572 allowed unauthenticated access to their SMS functions, and message histories showed the flaw being abused to send phishing texts since at least February 2022. I have covered what these engines see in what is Shodan.
Governments are saying it plainly
On 27 August 2026 the UK's National Cyber Security Centre warned of increased targeting of operational technology across multiple sectors, including the UK, with some limited real-world disruption. Its advice reads like a description of the typical public IP SIM install: make sure PLCs and HMIs are not directly exposed to the public internet, keep routers and remote access appliances within vendor support, change default credentials, and manage boundary devices only from a segregated management network. In February 2025 a group of international cyber agencies, with NCSC among those endorsing it, published joint guidance on securing exactly this class of edge device.
In the US, CISA's Binding Operational Directive 23-02 of June 2023 gives federal agencies 14 days to take an exposed management interface off the internet or put it behind zero trust controls, with zero trust the preferred option. It does not apply to UK businesses, but it is a clear statement of where good practice now sits.
Regulation is following. The EU Cyber Resilience Act obliged manufacturers to report actively exploited vulnerabilities and severe incidents from 11 September 2026, with a 24-hour early warning. Its main product requirements follow in December 2027. None of this bans a public IP on a SIM. But it puts manufacturers, integrators and customers under far more scrutiny when an exposed device is compromised.
And you may pay for the privilege
There is a quieter cost too. Every public address on the internet receives a constant background of scans and probes. Whether that traffic actually reaches your SIM depends on the operator's filtering, and whether you pay for it depends on the provider's terms. With some providers it clearly counts as data. Simbase's documentation is explicit: on a SIM with a public IP and an open firewall, inbound traffic is billed even when you did not ask for it, scanning and probes included. Academic work spotted the risk early. A 2014 study from the National University of Singapore showed that public IPs on 4G networks let an attacker drain a subscriber's data allowance and battery, because traffic is counted once it passes the operator's gateway. I have seen fixed IP SIMs on remote sites burn through allowances with nobody logged in, and it is one of the most confusing bills a customer can receive.
A public IP SIM costs more than its add-on fee. Add unsolicited data, the engineer time spent hardening and patching an internet-facing device, and the consequences if it is compromised. On that basis, most fleets are paying a premium for a risk they do not need.
None of this means a public IP can never be used safely. Sometimes it is the right answer, and I have set out how to do it properly in how to secure a public IP cellular router. The point is that it should be a deliberate decision, not the default.
From shortage to surplus?
Here is the idea I want to put forward, and I want to be honest that it is a prediction rather than a measured trend.
If security pressure pushes even a modest share of UK fleets off public IP SIMs and onto private addressing with outbound remote access, the public addresses those SIMs held go back into the pools of the operators and IoT providers that hold them. Providers can already reclaim them; Simbase, for example, stops its daily fee when a customer releases an address. Every address handed back increases the supplier's available pool. If addresses are released faster than new customers need them, that could reduce the need to buy or lease more IPv4. Whether it adds up to a genuine surplus depends on the balance between addresses returned and new demand.
What I expect that to mean in practice is not a flood of cheap IPv4 on the open market. Cloud, hosting and broadband demand dwarfs anything IoT will release, and suppliers may simply hold, reuse or sell the addresses. The effect, if it comes, will be local: public IPv4 on SIMs becoming easier to get for the customers who genuinely need it, and suppliers competing on secure remote access instead of on address rationing. The customer who moves wins twice, paying less and carrying less risk.
I have found no published data yet showing fleets migrating at scale, and I would treat anyone claiming a measurable IPv4 surplus from IoT with caution. But the incentives all point the same way, and in this industry, incentives usually win.
Seven alternatives to public IP SIMs
If most routers should not sit on a public IP, how do you reach them? The good news is that, in ordinary deployments, every option below gives remote access without the router accepting unsolicited connections from the public internet. That is not the same as being unreachable by every possible route. How a provider routes and filters traffic, who else shares an APN and what access tools the provider offers all matter, so the router still needs its own firewall.
There are still plenty of jobs where a public IP is the right call, from 4G or 5G backup into a business firewall and temporary sites waiting for a leased line, to VPNs that expect a fixed peer and legacy kit that cannot run a tunnel. I make that case, and explain how to run one safely, in why we still use public IP SIMs.
| Option | How it avoids a public SIM IP | Best for | Watch out for |
|---|---|---|---|
| Vendor cloud management | Router dials out to the vendor platform, which brokers access | Single-brand fleets, small teams | Vendor lock-in and per-device or data charges |
| Outbound VPN to a VPS hub | Routers dial a WireGuard or OpenVPN server you control | Mixed brands, fixed egress for partner whitelists | You run and patch the hub; build in redundancy |
| VPN back to head office | Routers dial the office firewall or concentrator | Sites that must join the corporate network | Office link becomes a dependency; segment the traffic |
| Zero trust overlay | Each device joins an identity-based overlay network outbound | Granular, per-user access to specific devices | Router support, licensing and relay performance |
| Private APN | Operator delivers SIM traffic privately into your network | Larger estates wanting operator-managed isolation | Setup cost and supplier dependence; still needs a firewall |
| Provider remote access | The IoT SIM provider brokers on-demand access | Occasional engineer access | Tied to that provider |
| IPv6 | Every device gets a global address without IPv4 | Greenfield fleets with IPv6-ready back ends | APN support and IPv6 firewall discipline |
Vendor cloud management
Most router vendors now offer a cloud platform the router connects out to. Teltonika's RMS is the one I see most in the UK. Management is paid in credits, one credit covering one device for 30 days, and its Connect and VPN remote access features draw on a data allowance, with a one-off free 5GB per company account. It is the quickest route off public IPs for a Teltonika fleet. The trade-off is that you are tied to one vendor's platform and pricing.
Outbound VPN to a VPS hub
My personal favourite for mixed fleets. Every router dials out to a small virtual server running WireGuard or OpenVPN, and engineers connect to the same server. The routers need no public IP, and the VPS gives you one fixed public address that partners can whitelist. That neatly solves the one requirement that most often justifies a public IP SIM. The catch is that you own the server, so patch it, monitor it and do not let it become a single point of failure. The pattern is covered in using a VPS for IoT and M2M.
VPN back to head office
Routers build tunnels to the office firewall or a VPN concentrator, so remote sites appear on the corporate network. It suits organisations whose IT team already runs a VPN estate. Just make sure remote sites land on their own segment, not the flat office LAN.
Zero trust overlays
Overlay networks such as ZeroTier and Tailscale give each device an identity and let you grant access device by device, without any inbound ports. ZeroTier's Scale plan, for example, is listed at $179 a month including 100 devices. Tailscale charges per user, with additional charges possible for tagged devices. Check router support and licensing carefully before committing a fleet. This is the model the cyber agencies now prefer.
Private APN
The operator delivers your SIM traffic privately into your own network, with private static addressing, so in a typical design devices are not directly reachable from the public internet. Designs vary: some private APNs carry everything back to your network, others include a controlled internet breakout, so ask exactly where traffic goes. It is the traditional enterprise answer, it is excellent at scale, and the trade-off in depth is in private vs public IP for cellular IoT. It is not automatically secure, though. Your own firewall still has to do its job.
Provider-side remote access
Some IoT connectivity providers now build access into the SIM service itself. Soracom's Napter, for example, gives on-demand remote access without a public IP on the SIM. It is convenient for occasional engineer access, but it ties you to that provider.
IPv6: the long game
IPv6 removes the shortage completely. Every device can have its own global address. It passed 50% of traffic to Google in March 2026, and EE has said its mobile network uses IPv6 with 464XLAT. Consumer mobile support is not the same as IoT support, though. IPv6 on IoT and M2M APNs varies by operator and provider, so ask before you design around it. The bigger catch is everything around the router. SCADA servers, VPN concentrators and many IoT platforms are still IPv4-only, and there is no NAT quietly blocking inbound traffic. The common mistake I see is engineers locking down IPv4 carefully, then enabling IPv6 on the mobile interface and leaving its firewall at a permissive default. If you turn IPv6 on, audit the IPv6 firewall with the same care. Buy IPv6-ready hardware now; it will still be in the field in 2035.
A worked example: 100 routers
Take a fleet of 100 cellular routers, all on public static IP SIMs, and compare two private alternatives. This is illustrative, not a quote, and the assumptions matter, so here they are.
- Address add-on prices come from Daisy's March 2024 tariff sheet: £7.50 public, £2.00 private. The sheet does not state the billing period; I have treated them as monthly, in line with its data tariffs.
- This compares selected recurring address and remote access costs only. Data charges, hardware, set-up fees, VAT and other support costs are left out. They can differ between architectures and belong in any real procurement comparison.
- Remote access platform costs vary widely, so I have used round placeholder figures, not supplier prices. The £100 a month management figure is an illustrative extra cost of running the private remote access solution, not the total cost of managing the fleet.
| Approach | Monthly cost (illustrative) | What you get | Exposure |
|---|---|---|---|
| Public static IP on every SIM | 100 x £7.50 = £750 | Direct inbound access to every router | High |
| Private static IP plus access platform | 100 x £2.00 = £200, plus about £100 platform and £100 extra management time = £400 | Fixed private addresses, brokered remote access | Low |
| Standard dynamic SIM plus outbound VPN | No address add-on, plus about £100 platform and £100 management = £200 | Routers dial out; no fixed address needed | Low |
On these assumptions only, including treating Daisy's add-ons as monthly, the private routes come out around £350 to £550 a month cheaper for 100 routers. Change the assumptions and the gap changes. A cheap public IP like Things Mobile's €2 a month narrows it sharply, and a heavily managed private APN can close it altogether. What the address fee does not show is the cost of the exposure itself: unsolicited data, hours spent hardening and patching internet-facing routers, and the bill if one is compromised. Run the numbers with your own supplier's prices. In my experience the result rarely favours public IP once those hidden costs are counted.
Is my SIM behind CGNAT? A two-minute check
- Read the router's mobile WAN address. It is on the status page of almost every cellular router.
- Look at the range. An address starting 10., 172.16. to 172.31., 192.168. or 100.64. to 100.127. is private or shared, so the router has no directly routable public IPv4 address on that interface. That usually means CGNAT, but it can also be a private APN or another provider-managed arrangement.
- Compare with what the internet sees. From a device behind the router, check your public address with any "what is my IP" service. If it differs from the WAN address, your traffic is leaving through a shared or provider-managed public address. Check that VPN routing on the router is not skewing the result.
- If they match, you probably have a public IP. Check your WAN firewall, confirm the admin interface is not reachable from the internet, and ask whether you really need it.
- Check IPv6 too. If the WAN shows an IPv6 address, review the IPv6 inbound firewall policy separately.
- Confirm with your provider. Ask your SIM supplier how the APN is configured before deciding exactly which architecture you are on.
What I would do now
- Audit every public IP SIM. List each one, what it is for, and whether it is still in use. Release the ones you do not need; with some providers a dormant public IP keeps billing.
- Default new deployments to private. Standard or private APN SIMs with outbound connectivity, and a public IP only when there is a written reason.
- Move remote access outbound. Vendor cloud, a VPS hub, a VPN to head office or a zero trust overlay. Routers that dial out survive network changes, failover and eSIM profile switches far better.
- Stop using the IP address as the identity. Authenticate devices with certificates or VPN keys, not source addresses.
- If a public IP stays, earn it. One exposed service, ideally a VPN, source whitelisting, admin interfaces off the WAN, strong credentials, monitoring and fast patching. My full checklist is in the cellular router security cornerstone.
- Plan for IPv6. Buy hardware with proper IPv6 firewalling and ask your SIM supplier and platform vendor for their IPv6 plans.
When that supplier warned me about IPv4 more than a decade ago, the advice was to design around public addresses. It was good advice then, for the wrong reason. Today the reason is security, and the reward for following it is a smaller bill, a smaller attack surface, and maybe, eventually, a few more public addresses left for the people who genuinely need them.
For how SIM routing and breakout affect where your data goes, see IoT SIM data sovereignty in the UK, and for addressing options on eSIM, the IoT eSIM guide. If you need a stable address rather than open inbound access, a fixed IP SIM on a private network is often the better fit.
Frequently asked questions
Do I need a public IP SIM to access my router remotely?
Usually not. A router on a standard or private APN SIM can dial out to a vendor cloud platform, a VPN server on a VPS, your office firewall or a zero trust overlay, and you reach it through that. A public IP still makes sense where a system must connect inbound to the router, whether from known sources such as a monitoring centre or from unpredictable ones, where legacy protocols or equipment cannot work through a tunnel, or where an outbound tunnel is impractical. Even then, restrict it to the sources and services that need it.
Do I pay for scanning traffic on a public IP SIM?
It depends on the provider, but often yes. Unsolicited traffic such as scans and probes can reach a SIM with a public IP, depending on the operator's filtering, and some providers state plainly that inbound traffic counts towards data usage. A strict firewall helps, but the safest fix is not to have a public IP unless you need one.
When did IPv4 addresses run out in Europe?
The RIPE NCC made its final /22 allocation from its free pool on 25 November 2019. Since then, recovered addresses go out only through a waiting list, one /24 per eligible new member, and everything else comes from the transfer market.
Is CGNAT a security feature for IoT routers?
Not by design. CGNAT stops unsolicited inbound connections from the internet, which hides routers from scanners, but it does nothing about compromised devices making outbound connections, attacks from inside the LAN, or other subscribers on the same network. Treat it as a useful side effect and still harden every device.
Why has IPv6 not replaced IPv4?
IPv6 is not backwards compatible with IPv4, so networks have to run both or translate between them. NAT and the IPv4 transfer market took away the urgency, and industrial systems are replaced slowly. IPv6 has passed half of traffic to Google and is strong on mobile networks, but many IoT platforms, private APNs and SCADA systems are still IPv4-only.
Can I keep a fixed public IP for outbound connections without one on every SIM?
Yes. Route your routers through a VPN hub on a VPS, or a provider gateway with a fixed outbound address, and partner systems see that single fixed address. The routers themselves stay on private addressing and, in a typical setup, are not directly reachable from the internet. This covers the most common reason people ask for public IP SIMs: a partner firewall that only accepts known source addresses.
Will Web3 or blockchain naming replace IP addresses?
No. Systems such as the Ethereum Name Service map names to identifiers, including IP addresses, so the traffic still travels over IPv4 or IPv6. They change who controls a name, not how packets are routed, and they create no new routable addresses.



