Intel introduced a set of knowledge middle updates at Computex 2026 in Taipei spanning compute, networking, and its AI accelerator roadmap. The headline merchandise is the introduction of Intel Xeon 6+ processors, paired with an expanded 800 Collection Ethernet lineup primarily based on Intel Ethernet E835 controllers and adapters. Intel additionally supplied an replace on its subsequent knowledge middle GPU, code-named Crescent Island. Collectively, the bulletins mirror Intel’s positioning round agentic AI, during which the CPU more and more serves because the management airplane for orchestration, concurrency, and knowledge motion throughout clusters.
Intel knowledge middle management framed AI scaling as a coordinated techniques downside moderately than a part improve cycle. As AI brokers grow to be extra autonomous and workloads grow to be extra distributed, Intel is emphasizing tight coupling between CPU, reminiscence, and networking to scale back bottlenecks and enhance effectivity below actual energy and rack constraints.
Xeon 6+ Extends the Xeon 6 Household with Density, Effectivity, and Scale-Out Focus
Xeon 6+ processors prolong the Xeon 6 household, emphasizing efficiency density, energy effectivity, and operational scale for cloud-native, network-intensive, and agentic AI-driven workloads. Intel mentioned Xeon 6+ is constructed on Intel 18A, marking the corporate’s first use of that course of node in an information middle CPU. The platform’s focus is sustained throughput in environments the place watts per rack, throughput per core, and predictable latency are main constraints.
Intel’s acknowledged function set targets scale-out infrastructure that should add new AI-adjacent providers with out requiring a disruptive knowledge middle redesign. Configuration highlights embody as much as 288 environment friendly cores, which Intel says ship as much as 2.5 instances extra efficiency than the earlier technology and as much as 45 p.c higher per-thread efficiency per watt versus the competitors, together with 12-channel DDR5 and 96 lanes of PCIe Gen5 with CXL help to maneuver knowledge throughout heterogeneous infrastructure. Intel additionally claims as much as 9:1 server consolidation versus 2nd Gen Xeon techniques. As well as, Intel is introducing Intel Software Power Telemetry on Xeon 6+, offering real-time, workload-level telemetry of CPU vitality and exercise to enhance visibility into consumption and utilization.
On safety, Intel highlighted silicon-level protections, together with Intel SGX and Intel TDX, aimed toward confidential computing and multi-tenant deployments. Intel additionally mentioned Xeon 6+ platforms are already being examined in telecom community infrastructures and configured into knowledge middle techniques throughout the ecosystem, with servers, networking, and built-in options from ASUS, Dell Applied sciences, Ericsson, GIGABYTE, HPE, Lenovo, Supermicro, and others creating on Xeon 6+ immediately.
Intel Ethernet E835 Targets Energy-Environment friendly 10GbE to 200GbE for AI and Virtualized Knowledge Facilities
Intel expanded its 800 Collection Ethernet portfolio with Intel Ethernet E835 controllers and community adapters, positioning networking as a key limiter for contemporary AI, cloud, and distributed workloads. The E835 line targets dense, virtualized deployments the place bandwidth and latency consistency have to be delivered inside tight energy envelopes.
Intel mentioned the E835 household helps as much as 200GbE and a variety of port configurations, together with 2x25GbE, 4x25GbE, 2x100GbE, and 1x200GbE, with further configurations enabled by Intel’s Ethernet Port Configuration Instrument. For effectivity, Intel mentioned the E835-CQDA2 adapter delivers as much as 1.9 instances increased efficiency per watt than the comparable NVIDIA ConnectX-6 Dx and 1.4 instances increased than the Broadcom BCM957508-P2100G, positioning the product line as lowering vitality consumption with out sacrificing throughput.
On offloads and knowledge path effectivity, E835 helps RDMA by way of RoCEv2 and iWARP, together with Dynamic System Personalization to streamline packet processing. For safety and manageability, Intel highlighted a {hardware} root of belief, signed SPDM, DMTF-based manageability, and OS help for Linux, VMware ESXi, and Home windows. Intel additionally known as out a ten+ yr lifecycle goal for long-term fleet standardization. Really useful pricing varies by configuration and is listed at intel.com/ethernet.
Xeon 6300 Provides a 12-Core Entry Possibility for SMB Servers
Intel additionally introduced basic availability of a brand new 12-core possibility within the Xeon 6300 household for entry servers, extending the platform past 8 cores. The important thing message is a drop-in improve path for current entry-level server designs, permitting SMB environments to extend compute capability with out a platform change, and with availability by main OEMs.
Crescent Island: Intel Updates Subsequent Knowledge Heart GPU for Inference and Token-Heavy Workloads
Intel supplied an replace on Crescent Island, its next-generation knowledge middle GPU constructed on the Xe 3P structure. Intel is positioning the half round reminiscence capability, bandwidth, and effectivity as differentiators for agentic AI inference, significantly as token-intensive workloads develop.
Based on Intel, Crescent Island pairs LPDDR5X with as much as 480GB of capability. It targets a 350W air-cooled PCIe kind issue for scale-out deployments the place rack-level thermals and energy supply are limiting elements. Intel additionally highlighted broad data-type help, starting from native FP4 and MXFP4 to FP64, in addition to expanded help for AI operations and scalability options. On software program, Intel reiterated its dedication to an open, programmable stack designed to scale back friction in heterogeneous environments, with Arc Professional cited as a improvement platform aligned with the identical Xe basis for ahead and backward compatibility.
Associated: Supermicro Launches 12 Xeon 6+ Optimized Platforms, Expands X14 and DCBBS Choices
In associated information, Tremendous Micro Laptop introduced 12 new server platforms optimized for Intel Xeon 6+ processors, spanning its Hyper, SuperBlade, FlexTwin, and GrandTwin households. Supermicro is emphasizing excessive core density and performance-per-watt for high-density cloud, virtualization, 5G analytics, and different throughput-intensive workloads. The brand new techniques broaden the corporate’s X14 lineup, with Supermicro positioning its DCBBS platform, a modular infrastructure strategy constructed from validated parts and subsystems, as the combination layer throughout the portfolio.
Supermicro’s up to date portfolio spans a number of server households tuned for various densities, cooling, and deployment fashions, giving operators a clearer path to standardize on a standard platform whereas nonetheless matching {hardware} to workload necessities. The Hyper collection targets mainstream rack deployments with single- and dual-socket 1U and 2U techniques that prioritize configurability. In follow, meaning versatile CPU choices, high-memory configurations for virtualization and knowledge providers, and the power to pair the platform with superior networking the place east-west site visitors and storage connectivity grow to be efficiency limiters.
For environments the place rack effectivity is the first constraint, Supermicro’s SuperBlade and multi-node designs push density increased whereas sustaining serviceability. SuperBlade packages as much as 10 compute nodes right into a compact 6U chassis, utilizing shared infrastructure to enhance utilization at scale, a match for giant fleet deployments that profit from simplified energy and administration domains. FlexTwin and GrandTwin take a multi-node strategy in rack kind elements, with FlexTwin emphasizing liquid-cooled, dual-socket nodes that function independently whereas sharing energy and cooling assets, and GrandTwin specializing in single-socket density optimized for high-core-count, E-core-heavy workloads the place throughput per rack and thermal effectivity matter.
Supermicro mentioned the brand new X14 platforms can scale as much as 576 environment friendly cores per server in dual-socket configurations, enhancing deployment effectivity and vitality consumption for large-scale knowledge facilities.
On the rack and knowledge middle stage, Supermicro positions DCBBS as the combination layer that ties these techniques collectively right into a modular AI infrastructure. Constructed from validated parts and subsystems, DCBBS is designed to scale back deployment friction by providing a repeatable building-block strategy that scales from particular person servers and networking to full rack-scale options, with supporting software program and providers for operators constructing out bigger cloud and AI footprints.
Supermicro additionally cited architectural positive factors it attributes to Xeon 6+ techniques, together with double the core depend, as much as 17 p.c increased IPC, 5 instances extra last-level cache, and 25 p.c quicker reminiscence help in comparison with earlier generations.
