Intel (NASDAQ: INTC) has unveiled its latest graphics processing unit, the Arc Pro B70, specifically designed for local AI workloads on workstations rather than for gaming or data centers. Priced at $949, the Arc Pro B70 offers a more affordable alternative to existing products from competitors like Nvidia and AMD, which retail at $1,800 and $1,299 respectively for similar models. This new GPU, boasting 32GB of RAM, positions Intel to potentially capitalize on a market shift where AI workloads move away from large-scale data centers.
The Arc Pro B70 features 32 Xe cores that deliver 22.9 TFLOPS of FP32 compute performance, along with 32GB of GDDR6 RAM that allows for a bandwidth of 608 GB/s. While this memory type does not match the high-bandwidth memory often found in data center GPUs, it is a cost-effective choice for smaller AI models. Intel claims the Arc Pro B70 can achieve 367 TOPS (trillions of operations per second), making it capable of handling a variety of AI tasks locally.
One significant advantage of the Arc Pro B70 is its ability to run larger context windows than Nvidia’s RTX Pro 4000 using the open-source Llama 3.1 8B model. Intel asserts that its new GPU offers faster response times, even with multiple concurrent users. This performance level could attract professionals in creative applications who are looking for cost-effective solutions. Additionally, users can link multiple GPUs to accommodate even larger AI models that may exceed a single GPU’s capacity.
The development of local AI processing is becoming increasingly relevant, particularly as companies seek to reduce reliance on mega data centers for routine AI tasks. For enterprise AI applications with predictable loads, workstations equipped with GPUs like the Arc Pro B70 could provide a viable, cost-effective alternative. This shift provides not only a financial advantage but also addresses privacy concerns associated with cloud-based processing.
Looking forward, as memory and computational power improve in workstations and standard PCs, it’s conceivable that more sophisticated AI models could process entirely on local machines. Intel’s advancements in AI capabilities are reflected in its latest PC chips, such as the Panther Lake laptop chip, which can deliver 180 TOPS of performance through an integrated AI accelerator. However, the ongoing memory shortage poses a challenge, and lower prices for memory chips are essential for broader adoption of advanced AI workloads on personal computers.
The Arc Pro B70 is a statement of Intel’s intent to be at the forefront of a potential shift towards local AI processing. While the company also targets the large data center market with its upcoming Crescent Island GPU, the Arc Pro B70 hints at a future where more powerful AI capabilities are accessible to individual users and small enterprises.
As investors consider Intel’s stock, it is worth noting that some analysts, including those from The Motley Fool, have identified other investment opportunities they believe may yield better future returns. These assessments often point to stocks that have performed exceptionally well historically, underscoring the competitive landscape in which Intel operates.
See also
Agile Robots Partners with Google DeepMind to Enhance Industrial Robotics Performance
DeepSeek Unveils Largest AI Model Yet, OpenAI Completes GPT-5.5 Development
Mistral Launches Voxtral TTS Model Supporting 9 Languages for Edge Devices
Dutch Court Orders xAI to Halt Non-Consensual Nude Image Generation, Fines Up to €100K Daily
