A Bitcoin user wants to receive a payment using Payjoin, a privacy-enhancing protocol that blends the sender’s and receiver’s inputs into a single transaction. They load Cake Wallet on their device, generate a Payjoin invoice, and share the URI with a sender. The sender’s wallet initiates the protocol handshake, but the connection times out. The user is behind a residential gateway or enterprise firewall without a public IP address—a condition known as Carrier-Grade NAT (CGNAT)—and the protocol cannot establish the inbound connection required for Payjoin to work. The transaction reverts to a standard payment, and the privacy benefit disappears.
This scenario reveals a fundamental constraint in Payjoin’s design that affects many real-world deployments, particularly for mobile and web-based wallets. Payjoin is not simply an optional privacy toggle; it is a bidirectional protocol that requires the receiver to act as a server accepting inbound connections from the sender. When network topology, firewall rules, or ISP infrastructure prevent that connection, the entire protocol fails—silently in many cases. Understanding why Cake Wallet web and other non-custodial implementations struggle with Payjoin connectivity is essential for anyone evaluating this privacy method or troubleshooting failed transactions.

How Payjoin differs from standard Bitcoin payments
A standard Bitcoin transaction has a simple broadcast model. The sender constructs a transaction, signs it with their private key, and broadcasts it to the network. Miners see the transaction, include it in a block, and the receiver’s wallet detects the confirmation. The receiver never needs to transmit anything to the sender; the payment is unidirectional and the receiver can be completely offline.
Payjoin inverts that model. The sender still initiates the payment, but instead of broadcasting immediately, they first contact the receiver’s Payjoin endpoint—a URL or server address embedded in a special invoice format—and propose the transaction. The receiver receives the proposal, examines it, and may add their own inputs to the transaction before returning a modified version. The sender then signs the receiver’s version and broadcasts the final result. Both parties’ inputs are now mixed, which makes chain analysis harder because observers cannot easily determine which inputs correspond to which party.
This privacy benefit comes at a cost: the receiver must be reachable and must respond quickly. The sender cannot complete the payment without hearing back from the receiver, and the receiver must have a way to accept inbound requests. For a user sitting behind a residential router connected to an ISP that uses Carrier-Grade NAT (CGNAT), or behind a corporate firewall that blocks inbound connections, that requirement becomes a dealbreaker. The receiver’s wallet cannot receive the inbound connection from the sender, the handshake fails, and the protocol falls back to a standard payment or times out entirely.
Carrier-Grade NAT and the inbound connection problem
Carrier-Grade NAT is a network architecture where an ISP overloads a single public IP address across thousands of customer devices. Instead of assigning each customer a unique public address, the ISP places customers behind a large shared NAT gateway. This solves IPv4 address exhaustion and reduces routing tables, but it creates a problem for any protocol that requires inbound connections: a sender cannot reach a receiver’s device directly because the device has no public IP address.
A device behind CGNAT can initiate outbound connections. It can request data from a server, which sends a response back along the same connection. But if an external party tries to contact the device first—by sending a packet to the device’s “address”—the ISP’s NAT gateway has no way to route that packet to the correct customer device. The gateway has no mapping in its translation table because the device never initiated the connection in the first place. The incoming packet is dropped, and the connection attempt times out.
Payjoin invoices typically contain a URL pointing to the receiver’s Payjoin endpoint. When a sender’s wallet processes the invoice, it sends an HTTP POST request to that URL, passing the proposed transaction for the receiver to evaluate. If the receiver is behind CGNAT and that URL points to an inbound port on the device (rather than a proxy or relay), the HTTP request will fail. The sender’s wallet waits for a response that never comes, times out after several seconds, and the payment falls back to standard broadcasting.
This is not a theoretical edge case. Residential broadband, many mobile carriers, and satellite internet providers use CGNAT or similar architectures. A user running Cake Wallet on an Android phone connected to a carrier network, or on a desktop behind a home router on a CGNAT-enabled ISP, will likely encounter this problem if they attempt to receive a Payjoin payment. The protocol was designed with the assumption of a stable, publicly routable network address—an assumption that no longer holds for a growing fraction of internet users.
Firewall blocking and enterprise network constraints
Corporate and institutional networks impose additional inbound connection restrictions. A user attempting to run Cake Wallet at an office with a managed firewall will find that inbound HTTP or custom protocol connections are blocked by default. Even if the user has a public IP (through a VPN endpoint or direct office IP), the corporate firewall policy may prohibit inbound connections on arbitrary ports. A payment processor or business might want to receive Payjoin payments from customers but cannot do so from behind a restrictive firewall without explicit security approval or infrastructure changes.
Some networks block not just inbound connections but also outbound connections to non-standard ports. A Payjoin endpoint might listen on port 8080 or 9000, but the corporate firewall policy may restrict outbound to ports 80 and 443 only. A sender’s device would attempt to connect on the Payjoin port and time out. Even if the receiver is technically reachable, the network policy makes them unreachable for the purposes of Payjoin.
The severity of this constraint depends on the deployment context. For an exchange or payment processor operating a server in a datacenter with a static public IP and no firewall restrictions, Payjoin is more viable. For an individual using a wallet on a personal device behind a home router or mobile network, the probability that an inbound connection succeeds is much lower. Cake Wallet web and similar non-custodial wallets do not control the network they run on; they expose protocol constraints rather than solving them.
Why relay servers and proxy architectures complicate the solution
A natural response to CGNAT and firewall problems is to use a relay or proxy server. Instead of embedding a direct device address in the Payjoin invoice, the invoice points to a relay server that acts as an intermediary. The receiver’s wallet contacts the relay when it comes online, informing it of an inbound address for Payjoin requests. A sender then sends their Payjoin proposal to the relay, which forwards it to the receiver. The receiver responds through the relay, which sends the response back to the sender.
This architecture solves the connectivity problem but introduces trust and custody concerns. A relay server is a centralized intermediary that can observe Payjoin proposals, responses, and the timing of payments. It can log IP addresses, infer transaction amounts from message sizes, and potentially link senders and receivers. For a privacy-conscious user, a relay defeats much of the purpose. Additionally, if the relay becomes unavailable, times out, or is overloaded, Payjoin fails again. The user has traded one form of unreliability for another while adding a new privacy vulnerability.
Some Payjoin implementations use distributed relay systems or allow users to run their own relay, but this introduces operational complexity. A user would need to configure their wallet to use a relay endpoint, understand that the relay is not a trusted service, and decide whether the privacy trade-off is acceptable. For most users, this is too much friction. Cake Wallet web and other consumer-focused wallets do not typically include relay configuration options, which limits their Payjoin usability to users with direct public IP connectivity.
There is also a custody distinction worth noting. A custodial relay—one operated by an exchange or payment processor—holds the inbound endpoint and thus can technically intercept or modify Payjoin proposals before they reach the receiver. A non-custodial relay operated by a third party can still observe metadata and timing. A relay operated by the user themselves requires infrastructure knowledge and ongoing maintenance. Each option involves compromises that undermine Payjoin’s original privacy goal to some degree.
Timeout behavior and fallback mechanisms in Cake Wallet web
When a Payjoin request times out, a wallet must decide what to do. Some wallets discard the transaction entirely and require the user to start over. Others automatically fall back to a standard Bitcoin payment. The behavior depends on the wallet’s design and the user’s configuration. cake wallet web and similar non-custodial wallets typically implement a timeout window of 10 to 60 seconds; if the Payjoin endpoint does not respond within that window, the wallet broadcasts a standard transaction instead.
This fallback is convenient for users—they do not experience a complete payment failure—but it is also deceptive. A user who requested a Payjoin transaction because they valued the privacy benefit receives no warning that their payment has fallen back to standard broadcasting. The transaction succeeds from the user’s perspective, but without the privacy properties they expected. An observer analyzing the blockchain sees a normal payment, not knowing that an enhanced transaction was intended but failed due to network connectivity.
Better wallet implementations display an explicit warning if Payjoin times out and the transaction reverts to standard broadcasting. They may also log the timeout and allow the user to retry or adjust settings. Some wallets increase the timeout window for mobile or unreliable networks, accepting a longer wait time in exchange for a higher success rate. The tradeoff is that a longer timeout makes the payment experience slower and less predictable.
The root issue remains unchanged: the protocol assumes bidirectional connectivity that many users do not have. A fallback mechanism is a usability patch, not a solution to the underlying architecture constraint. As long as Payjoin requires inbound connections, users behind CGNAT or firewalls will experience degraded reliability, whether they are aware of it or not.
Why Payjoin remains impractical for most mobile and home users
Mobile devices are particularly constrained. An Android phone connected to a cellular network is almost certainly behind CGNAT; the carrier assigns one public IP to thousands of devices. The phone can make outbound requests, but it has no public, routable inbound address. A user running Cake Wallet on that device and attempting to receive a Payjoin payment will fail unless they use a relay or connect to a different network. Even on Wi-Fi, a home router typically does not allow inbound connections to arbitrary ports by default. Port forwarding can be enabled manually, but that requires technical knowledge and creates security implications.
Desktop users with static public IPs and permissive firewalls are rarer. They may be operators of payment processors, merchants with server infrastructure, or users who have explicitly configured their networks for inbound access. For a casual Bitcoin user, receiving Payjoin payments is not practical without additional infrastructure or significant network reconfiguration.
The asymmetry is also important to understand: a sender can typically broadcast a standard Bitcoin transaction without any special connectivity requirements. They initiate outbound connections to the Bitcoin network, which is all that is needed. A receiver, by contrast, needs to be reachable inbound. This means Payjoin is a feature that benefits senders more than receivers—senders get better privacy by mixing inputs with a cooperative receiver—while the receiver bears most of the connectivity burden. A receiver might be willing to accept this tradeoff for a business context, but for individual users, the incentive structure does not align.
Alternative privacy approaches and their practical advantages
Bitcoin Silent Payments, specified in BIP-352, offer a different approach to receiver privacy. Instead of mixing inputs (which requires coordination), Silent Payments use elliptic curve mathematics to allow a sender to derive a unique, non-reusable address for each payment without any communication with the receiver. The receiver remains completely offline; they scan the blockchain to detect payments sent to them. This eliminates the inbound connectivity requirement entirely.
Silent Payments trade computational cost for connectivity: scanning requires more local work on the receiver’s device, but the protocol is unidirectional. A user can receive Silent Payments regardless of their network topology, firewall configuration, or whether they are online at the moment the payment is made. Cake Wallet web and other implementations increasingly support Silent Payments, and this represents a practical shift toward privacy methods that do not require bidirectional connectivity.
Another approach is to accept that privacy requires a custodial intermediary in some contexts. A merchant or user who regularly receives payments might run a dedicated Payjoin server in a datacenter, with a static public IP and no network restrictions. The receiver’s wallet points to that server, which handles inbound connections and forwards proposals to the receiver through a separate channel. This separates the wallet from the inbound endpoint, but it introduces custody and operational complexity.
Litecoin’s MWEB (Mimblewimble Extension Block) layer, documented on the Litecoin Project website, offers another alternative: privacy at the protocol level rather than through transaction-construction techniques. Instead of mixing inputs, MWEB obscures transaction amounts and receiver addresses at the protocol layer. This approach shifts privacy to the base layer and does not require coordination between sender and receiver.
Protocol design lessons and future wallet development
Payjoin’s timeout and connectivity failures highlight a broader lesson: privacy-enhancing protocols must account for real-world network topology, not just idealized assumptions. A protocol that works seamlessly in a datacenter or on a research testbed may fail for the majority of users who lack public IP addresses. Wallet developers face a choice: implement features that only work for a minority of users, implement relays that compromise privacy, or migrate to protocols that do not require inbound connectivity.
The future of Bitcoin privacy at the transaction layer likely favors approaches that are unidirectional or that do not require real-time coordination. Silent Payments, protocol-level privacy improvements like MWEB, and onchain privacy tools that do not require receiver participation are more aligned with how users actually connect to networks. Payjoin may remain useful for businesses and sophisticated users with controlled network environments, but for casual Bitcoin users, it is increasingly impractical.
Wallet developers and users evaluating privacy tools should understand the connectivity assumptions buried in protocol specifications. A feature that works in demonstration fails in deployment when real-world constraints are introduced. Testing on different network types—residential CGNAT, mobile carriers, corporate firewalls, VPNs—reveals gaps that do not appear in documentation. As non-custodial wallets like Cake Wallet web mature, the evaluation of privacy features should include not just what they promise, but where and for whom they actually work.
Frequently asked questions
Why do Payjoin payments fail when I use cake wallet web from a home network or mobile device?
Most home networks and all mobile carriers use Carrier-Grade NAT (CGNAT), which assigns one public IP address to thousands of users. Payjoin requires the receiver’s wallet to accept an inbound connection from the sender, but behind CGNAT you have no public, inbound-routable address. The sender’s wallet times out waiting for a response, and the payment falls back to a standard transaction without Payjoin’s privacy benefit.
Can I use a relay server to fix Payjoin timeouts?
A relay server can forward Payjoin proposals between sender and receiver, solving the connectivity problem but introducing a centralized intermediary that can observe transaction metadata and timing. Some wallets and implementations support relays, but this requires explicit configuration and the relay itself becomes a privacy and availability dependency. For many users, Silent Payments or other unidirectional privacy methods are more practical.
Is cake wallet web Payjoin support useful if I am behind a firewall?
Payjoin will time out and fall back to standard transactions unless your network has inbound connectivity on the endpoint port. If your firewall blocks inbound connections (as most corporate and home firewalls do by default), Payjoin is not practical without manual port forwarding or a relay. For home users and most mobile users, Silent Payments or protocol-level privacy methods are more reliable alternatives.