Home / Technology / Nvidia's High Prices Push AI to Intel
Nvidia's High Prices Push AI to Intel
2 Jul
Summary
- Intel's Arc Pro B70 offers 32GB VRAM for around $1,000.
- Nvidia's RTX 5090, with 32GB memory, costs up to $4,000.
- Intel's GPU targets AI workloads where VRAM is crucial.
The high cost of Nvidia's latest consumer GPU, the RTX 5090, is driving AI developers toward alternatives, particularly Intel's Arc Pro B70. Launched in 2026, the RTX 5090 features 32GB of GDDR7 memory but retails for up to double its $2,000 MSRP, reaching $4,000. This situation creates a 'VRAM crisis' for local AI needs. Intel's Arc Pro B70, priced around $1,000, offers a comparable 32GB of GDDR6 memory. This positions it as a cost-effective choice for AI workloads requiring substantial VRAM.
The Arc Pro B70, utilizing Intel's BMG-G31 chip, focuses on professional AI tasks rather than gaming. While Nvidia and AMD are concentrating on profit maximization with their high-end offerings, Intel is addressing a gap in the local AI market. Testing indicates that four Arc Pro B70s can rival the RTX 5090 in certain decode tasks due to superior memory bandwidth, though the RTX 5090 retains an advantage in compute-intensive scenarios. However, for memory-bound AI models, the Arc Pro B70's configuration offers more memory than a comparable RTX 5090 setup at a lower cost.
Despite its advantages, the Arc Pro B70 faces challenges in ecosystem adoption. Nvidia benefits from superior software support, including its CUDA-based software stacks and optimized libraries, which Intel's hardware cannot directly emulate. Intel's own software initiatives are improving but are generally considered less mature than AMD's ROCm stack, which itself lags behind Nvidia's established ecosystem. Nevertheless, the Arc Pro B70's availability close to its MSRP and its substantial VRAM make it a formidable and scalable option for AI professionals seeking an alternative to Nvidia's premium GPUs.
