Intel Crescent Island Squeezes 1.3 PFLOPs Of FP8 Into Just 350W As D270 Targets The Competition With Value-Oriented LPDDR5X Up To 480 GB
Intel has formally introduced its new GPU D-Series, branding its previous Crescent Island architecture as the Intel GPU D270. Designed to compete in the artificial intelligence sector, the hardware packs 1.3 peak PFLOPS of FP8 compute power into a 350W thermal envelope. The accelerator features 1.5 TB/s of memory bandwidth and incorporates up to 480 GB of value-oriented LPDDR5X memory. This configuration specifically targets enterprise and agentic artificial intelligence inference workloads by offering high capacity at a competitive power profile.
Listen to Live Briefing
Real-time synthesized voice briefing · Live Feeds Desk
- ✓ Intel officially named the Crescent Island accelerator the GPU D-Series on October 12, 2026.
- ✓ The Intel GPU D270 delivers 1.3 peak PFLOPS of FP8 compute power.
- ✓ The accelerator operates at a 350W power limit.
- ✓ The hardware supports up to 480 GB of LPDDR5X memory.
What changed
Intel officially named the Crescent Island accelerator the GPU D-Series, detailing its complete specifications including 1.3 PFLOPS of FP8 compute and up to 480 GB of LPDDR5X memory.
Live updates
-
Intel Unveils D270 AI Accelerator With 1.3 PFLOPS And 480 GB LPDDR5X
Intel has formally introduced its new GPU D-Series, branding its previous Crescent Island architecture as the Intel GPU D270. Designed to compete in the artificial intelligence sector, the hardware packs 1.3 peak PFLOPS of FP8 compute power into a 350W thermal envelope. The accelerator features 1.5 TB/s of memory bandwidth and incorporates up to 480 GB of value-oriented LPDDR5X memory. This configuration specifically targets enterprise and agentic artificial intelligence inference workloads by offering high capacity at a competitive power profile.
Why it matters
The rollout of the GPU D-Series represents an effort by Intel to capture a larger share of the enterprise artificial intelligence infrastructure market against competitors like Nvidia. By leveraging cost-effective memory solutions such as LPDDR5X alongside high-density computing, the company aims to provide scalable hardware for cloud and data center clients. This product line follows other recent architectural milestones, including the rollout of E-core-only Clearwater Forest processors for servers.
What is confirmed
- Intel officially named the Crescent Island accelerator the GPU D-Series on October 12, 2026.
- The Intel GPU D270 delivers 1.3 peak PFLOPS of FP8 compute power.
- The accelerator operates at a 350W power limit.
- The hardware supports up to 480 GB of LPDDR5X memory.
- The device provides 1.5 TB/s of memory bandwidth.
Still unconfirmed
- Intel has integrated two Crescent Island GPUs onto a single liquid-cooled OAM form factor card.
What to watch next
- Deployment timelines and pricing details for enterprise customers adopting the GPU D270.
- Official benchmarks measuring the D270 performance against competing artificial intelligence accelerators from Nvidia.
confidence 100%Sources used for this update (21)
- stockanalysis.com — Intel (INTC) Stock Price & Overview
- www.intel.com — Intel Inside - Built for AI
- www.intel.com — Download Intel Drivers and Software
- en.wikipedia.org — Intel - Wikipedia
- finance.yahoo.com — Intel Corporation (INTC) Stock Price, News, Quote & History ...
- www.intel.ie — Intel Company Overview and Future of Technology | Intel®
- en.wikipedia.org — List of Intel processors - Wikipedia
- intel.wd1.myworkdayjobs.com — Intel Careers
- builders.intel.com — Hardware - Intel® Industry Solution Builders
- Wccftech — Intel Crescent Island Squeezes 1.3 PFLOPs Of FP8 Into Just 350W As D270 Targets The Competition With Value-Oriented LPDDR5X Up To 480 GB
- Yahoo Finance — Intel aims to boost AI turnaround with new GPU chips to take on Nvidia
- Intel — Intel Accelerates Open Infrastructure for the Agentic AI Era
Community Sentiment: How do you assess this situation?
Voice your perspective · Real-time aggregated sentiment from the Live Feeds community