Nvidia’s RTX 5090 is a uniquely capable GPU – the company’s most high-end consumer GPU to date features 21,760 CUDA cores and 32GB of GDDR7 memory in its scaled-down GB202 configuration.
It also happens to be one of the most expensive GPUs consumers can buy at retail in 2026, going up to twice its suggested retail price of $2,000 for the Founders Edition at launch, forcing many cost-sensitive AI builders to turn to older Nvidia GPUs equipped with 24GB of VRAM, such as the RTX 4090, 3090Ti and 3090.
As AI models continue to advance in complexity (and memory requirements), a DRAM crisis makes it impossible for many to obtain a GPU capable of meeting their local AI needs, although one option seems largely overlooked: Intel’s Arc Pro B70.
A mid-range GPU with tons of VRAM
Most consumers don’t think of Intel as a GPU maker, but instead associate it with consumer and professional processors. This is despite the fact that the chipmaker offers some of the best value for money options in the mid-range segment in which it currently competes.
Intel’s Arc Pro offerings, however, are decidedly different; They focus exclusively on professional-grade, or rather AI-centric, workloads, while ignoring any pretense of meeting the needs of gaming consumers, even as they remain capable of running most titles that can be thrown at them.
The Intel Arc Pro B70 is its high-end offering and costs a reference price of $950, with most retailers and OEM partners selling SKUs for around $1,000. With 32GB of GDDR6 memory and a price that’s effectively a quarter of most RTX 5090 SKUs on sale at the time of writing, it presents itself as a value-focused alternative to Nvidia’s Blackwell-based giant.
It leverages the Intel BMG-G31 “Big Battlemage” chip that was originally planned to be installed on its since-canceled Arc B770 GPU, as it aims to fill a void in the local AI space that AMD and Nvidia are unwilling to fill given their focus on maximizing profits at the higher end of the spectrum with enterprise customers.
For those interested in configuring 4 of the Arc Pro B70, Puget Systems went ahead and performed the necessary testing, even drawing a direct comparison with the RTX 5090, which it found to be 4-5 times faster in decoding tasks thanks to a significantly greater amount of in-game memory bandwidth (1792 GB/s versus 608 GB/s).
The observations also highlighted a slightly obvious caveat in the comparisons: models that require more compute and bandwidth relative to memory will look to the RTX 5090, which offers a significant advantage even over multiple Arc Pro B70 GPUs, but those that require a large amount of memory for their settings would find a much more cost-effective B70 configuration with access to more memory than a corresponding RTX 5090-based offering.
Nvidia’s value proposition, however, goes deeper than the silicon itself, and that’s where Intel is struggling to find a willing buyer, thanks to better software support on various operating systems, CUDA-based software stacks that the Arc Pro can’t emulate, and Nvidia-coded libraries that won’t run on Intel hardware.
Intel’s own offerings (oneAPI, OpenVINO, and IPEX) are improving but are widely considered to lag behind even AMD’s ROCm stack, which in turn lags behind Nvidia’s ecosystem. Despite this, the Arc Pro B70, as noted above, benefits from the lack of a true high-VRAM alternative and remains available close to its stated MSRP, making it a formidable and future-proof alternative to Nvidia’s heavyweight GPU.
Follow TechRadar on Google News And add us as your favorite source to get our news, reviews and expert opinions in your feeds.
