Trending Articles
Trending Topics
Trending
As we all wait to see the shake-out of Nvidia’s partnership with Intel – what will it mean for Intel’s GPU family, from laptops to data centres? – the chip giant announced a new entry in its AI accelerator portfolio: Crescent Island.
This data centre GPU has been designed, Intel says, to meets the twin needs of growing AI inference workloads and energy efficiency.
“AI is shifting from static training to real-time, everywhere inference – driven by agentic AI,” said Sachin Katti, Intel CTO.
“Scaling these complex workloads requires heterogeneous systems that match the right silicon to the right task, powered by an open software stack. Intel’s Xe architecture data centre GPU will provide the efficient headroom customers need – and more value – as token volumes surge.”
Crescent Island is aimed at delivering high performance-per-watt and Intel claims it is ideal for “tokens-as-a-service providers” and “inference use cases”.
Intel hasn’t yet released any details of when Crescent Island will be released.
The announcement was made at the 2025 OCP Global Summit, with OCP standing for Open Compute Project. Intel will also showcase a new rack design built around its Gaudi 3 AI accelerator.
Intel continues to stick to its position that it is “uniquely positioned” to deliver solutions to meet the growth of “everywhere inference”.
To quote its press release, “success depends on more than powerful chips – it requires systems-level innovation. From hardware to orchestration, inference needs a workload-centric, open approach that integrates diverse compute types with an open, developer-first software stack – delivered as systems that are easy to deploy and scale.”
So why choose Intel? “Intel is uniquely positioned to deliver this end-to-end – from the AI PC to the data centre and industrial edge – with solutions built on Intel Xeon 6 processors and Intel GPUs.”
There is an argument there, and last week I attended the launch of Intel’s Clearwater Forest chips – now known as Xeon 6+ E-core – where it talked convincingly for the need of such efficient processors to meet the needs of cloud providers and telcos in particular.
With talk of an AI bubble, perhaps some AI service providers will be drawn to an offering that is both affordable and power-efficient.
Intel will also be showing a rack implementation of its Gaudi 3 accelerators for the first time. The promise here is easy scaling, with the design incorporating liquid cooling along with support for high memory bandwidth domain (HBD).
However, IBM aside for Gaudi 3’s launch announcement, we await any big name partnerships for Intel’s AI accelerator.
