RTX 5070 allegedly delayed until early March to counter AMD RX 9070 launch

A new rumor suggests that Nvidia's mid-range RTX 5070, previously slated for a launch later this month, has been delayed to early March, per MEGAsizeGPU at X. Its elder Ti-class sibling should still arrive on time. However, considering the infamous RTX 5090/RTX 5080 launch, we cannot comment on GPU's retail availability. If true, it's probably no coincidence that this "delay" coincides with AMD's Radeon RX 9070 series.

To recap, the RTX 5070 was unveiled by Jensen Huang at CES last month, wielding the GB205 GPU at its heart with 48 SMs (6,144 CUDA cores) and 12GB of GDDR7 memory. The GPU offers a 192-bit interface, equating to six memory modules, filled by 28 Gbps GDDR7 ICs for 672 GB/s of bandwidth. Nvidia has set a launch MSRP of $549 for the RTX 5070, which is $50 cheaper than the last-generation RTX 4070. Instead of declaring a definite launch date, Nvidia said the RTX 5070 would supposedly launch with the RTX 5070 Ti in February, but that's no longer true.

Renowned tipster MEGAsizeGPU, in a tweet, alleged that the RTX 5070 is delayed, with retail availability pushed back to early March, likely to thwart AMD's RDNA 4 launch. In light of recent leaked benchmarks putting the RX 9070 XT just behind the RTX 4080 Super (in raster), Nvidia's RTX 5070 marketing may hinge on pricing instead of performance.

During the RTX 50 reveal, Jensen Huang very proudly asserted the RTX 5070 is equal to the RTX 4090 (with MFG), which might be misleading for many unsuspecting consumers. The RTX 5080 barely edges the RTX 4080 Super by just 9%, which is quite disappointing for a gen-on-gen uplift, especially considering the RTX 4080 was 38% faster than the RTX 3080 Ti. In terms of other capabilities, Blackwell offers many new features such as Reflex 2.0, Smooth Motion, Multi Frame Generation, and.... more AI horsepower. Still, it's best to temper expectations and anticipate a 15-20% uplift in (raster) performance, at best, versus the RTX 4070.

Both cards in the RX 9070 family are reportedly packed with 16GB of memory, and as a result, AMD is marketing these GPUs as 4K-ready. The same cannot be said about the RTX 5070, which might struggle with VRAM-hungry titles at higher resolutions. Even though Nvidia is working on Neural Texture Compression (NTC) technology to reduce texture memory footprint, it's still a ways off from mainstream adoption.