The Raspberry Pi Zero 3 W and Pi 4 Model A surface in the Linux kernel tree
On 27 October 2026, Device Tree files for a Raspberry Pi Zero 3 W and a Raspberry Pi 4 Model A were merged into Raspberry Pi’s official Linux kernel tree. The Zero 3 W would run the BCM2711 SoC, the Pi 4’s chip: the Zero form factor is finally moving to a 64-bit quad-core.
27 October 2026. Device Tree patches for a Raspberry Pi Zero 3 W and a Raspberry Pi 4 Model A are merged into Raspberry Pi’s official Linux kernel tree. 27 October 2026. The Device Tree file for the Zero 3 W ties it to the Broadcom BCM2711, the same quad-core Cortex-A72 SoC used by the Pi 4. Why it matters: five years after the Zero 2 W, the Zero form factor is about to jump two CPU generations at once — and to reprice what a very cheap, always-on self-hosted node can do.
The Zero 2 W is five years old, and the form factor needs a successor
The Raspberry Pi Zero 2 W shipped in October 2021. It carried an RP3A0 (a BCM2710A1) — four 64-bit Cortex-A53 cores at 1 GHz. That was progress over the original Zero, but the base was still the Pi 3’s: a generation behind the full-size boards. Since then Raspberry Pi has delivered the Pi 5 (quad-core Cortex-A76) and the component market has changed shape — rising RAM prices, supply bottlenecks, and cost pressure that has already pushed the Pi 5 2 GB to $77.50 and the Pi 4 2 GB above $67.
That is the context in which the Device Tree merge makes sense. The fact that a Zero 3 W is defined around the BCM2711 — the Pi 4’s quad-core Cortex-A72 — suggests Raspberry Pi wants to put real compute back into the smallest board in its catalogue, without reaching for the Pi 5’s silicon. For a self-hoster the difference is concrete: the BCM2711’s A72 cores are markedly faster than the Zero 2 W’s A53 cores, on a board that keeps the Zero’s size, power draw and price.
The Pi 4 Model A, the other surprise in the batch
The second discovery is a Raspberry Pi 4 Model A. The Model A line is the cut-down, cheaper take on the B boards: smaller, lower power, fewer I/O ports and no Ethernet, but at a lower price. A Pi 4 Model A would repeat that recipe on the Pi 4 base — a way to offer a quad-core A72 at a tight price, without the features some use cases do not need.
Michael Larabel of Phoronix, who spotted the merge, calls the strategy “logical” in a market where components keep getting pricier: give a cheaper option to those who do not need a full-featured, current-generation SBC. The corresponding pull request is visible on Raspberry Pi’s kernel repository — github.com/raspberrypi/linux/pull/7711.
A point of method is worth underlining: this is a merge into the kernel tree, not an official announcement. Raspberry Pi has said nothing about price, date, RAM capacity or availability. The only thing the Device Tree confirms is the Zero 3 W’s SoC. Everything else is reading signals.
What this changes for a self-hosted fleet
For anyone running services at home, the promise of a Zero 3 W is specific. The Zero 2 W is already the default for light, always-on nodes: Pi-hole, secondary DNS, an Uptime Kuma probe, sensors, a small WireGuard VPN, an MQTT relay. Its limit was never size or power — it was the CPU. Moving from four A53 to four A72 cores changes what a node of this size can reasonably run: more headroom for network filtering, background jobs, compression, even small containerised services.
The table below summarises the generational gap, from the Zero 2 W’s public specs and what the Device Tree confirms for the Zero 3 W.
| Specification | Zero 2 W (2021) | Zero 3 W (upcoming, per the DT) |
|---|---|---|
| SoC | BCM2710A1 (RP3A0) | BCM2711 |
| CPU | 4× Cortex-A53 @ 1 GHz | 4× Cortex-A72 (clock TBC) |
| Architecture | 64-bit | 64-bit |
| RAM | 512 MB | TBC |
| Connectivity | 2.4 GHz Wi-Fi + BT 4.2 | TBC |
Caution is warranted on RAM: more than the CPU, it will decide what a Zero 3 W can actually absorb. If Raspberry Pi keeps 512 MB, the gain stays comfortable but bounded; if the board moves to 1 GB or more, it becomes a small service machine in its own right.
Then there is the energy argument, easy to overlook. A node that runs 24/7 only pays for itself if its power draw stays in the single digits — that is exactly what the Zero form factor exists for, running on watts rather than tens of watts. A Zero 3 W on the BCM2711 would keep that structural advantage while unlocking more CPU, so the most useful homelab scenario is a permanent node that can finally absorb heavier ad-hoc tasks on top of its base load without moving the electricity bill.
The low-cost alternatives — Orange Pi, NanoPi and assorted clones — do exist, but none enjoys the software ecosystem and kernel stewardship of Raspberry Pi. For hardware meant to run for years without intervention, that ecosystem, more than the silicon, is the decisive argument.
One open question is whether a Zero 3 W inherits the Zero 2 W’s 512 MB of RAM or steps up. On a BCM2711, RAM becomes the binding constraint far sooner than CPU: a 1 GB Zero 3 W could run containerised services that a 512 MB board chokes on, while still sipping power. That single specification, more than the SoC itself, will decide whether the board is a drop-in Pi-hole upgrade or a genuine little server.
Verdict
If you need to buy light nodes soon, wait for the official Zero 3 W announcement before stacking up Zero 2 W boards: a move to A72 cores on this form factor would durably change the power-to-price ratio, and buying the old generation a few weeks before a probable successor would be poor timing. If you already run a fleet of Zero 2 W boards, change nothing: they remain perfectly suited to the light loads they carry, and a switch is only justified where the CPU is the bottleneck. If you are speculating about specs, hold yourself to the one verified fact — the Device Tree confirms the BCM2711, nothing more — and watch for the official announcement: no price, RAM or date has been published, and Raspberry Pi can still adjust the sheet before launch. What the kernel already says is that the Zero form factor will not sit still for another five years.