Could a handheld equal the performance of a home console? Leaked new specs for Sony’s next-generation PlayStation 6 handheld imply that it could not only equal but, in some cases, outperform the entry PlayStation 5 and leave rivals such as the Xbox ROG Ally X behind.
At its core is the Canis APU, a purpose-built, monolithic 135mm² die built on TSMC’s advanced 3nm process. It features an asymmetric six-core CPU configuration of four high-performance Zen 6c cores and two Zen 6 Low Power cores for operating system and background activities. This design keeps up to 20% more CPU performance available for gaming workloads, an essential feature in a power-limited handheld environment.
Graphics processing falls to a 16 Compute Unit RDNA 5 GPU, running at around 1.20 GHz on the go and 1.65 GHz when docked. RDNA 5’s architectural improvements are significant: per Compute Unit rasterization performance is 40–50% higher than RDNA 3.5, with even more lift in ray tracing. These gains are due to a mix of IPC gains of about 10% above RDNA 4 and added bandwidth, as the PS6 portable’s 192-bit LPDDR5X-8533 memory interface provides a 60% more bandwidth than the Ally X. This memory system supports up to 48 GB of RAM, though configurations are unknown.
The RAM capacity question is not merely theoretical. Developers who were consulted on the leak emphasized that24–36 GB will be necessary to enable next-generation features fully, especially AI-powered systems and memory-hungry rendering techniques such as Unreal Engine 5’s Nanite. These kinds of features require big, speedy memory pools to prevent asset streaming and geometry processing bottlenecks.
Performance estimates from Moore’s Law is Dead position docked rasterization output at 55–75% of a PS5, but with ray tracing performance between 1.3× and 2.6× above getting close to the PS5 Pro’s level. With AMD’s FSR 4 upscaling, patched games would run at close-to-PS5 fidelity, and unpatched titles would run at lower settings. This fits into general industry trends wherein AI-accelerated upscaling and frame generation balance raw rasterization shortfalls in mobile-class GPUs.
The jump from RDNA 3.5 to RDNA 5 is especially pertinent here. As demonstrated in recent RDNA 3.5 APU benchmarks, architectural improvements such as doubled texture sampling rate for generic operations, better memory compression, and LPDDR5X-optimized controllers provide tremendous gains in low-power conditions. RDNA 5 expands on this base with more thorough compute and ray tracing pipelines that scale efficiently within the handheld’s thermal envelope.
Backward compatibility with PS4 and PS5 games is assured, but no mention is made of PS3 backward compatibility. Hardware specs involve a MicroSD card slot, M.2 expansion port, haptic feedback, dual microphones, and a touchscreen reflecting Sony’s desire to marry console-level performance with portable convenience.
Cost estimates are also interesting. With better 3nm yields and falling RAM costs, estimates put production costs low enough for Sony to actually sell the device at $399–$499 without taking the usual launch loss. If true, this would be a very unusual case of break-even launch for Sony hardware, albeit extremely slim margins.
The handheld PlayStation 6 should come after the flagship PS6 console, probably in 2027 or beyond. If the leaked spec proves true, it will be a remarkable engineering feat: shrinking a console-grade Zen 6 + RDNA 5 design into a sub-15 W handheld package and providing ray tracing performance comparable to fixed systems. That would be a generation step in an industry where portable has perennially trailed behind living room gear.
