The year 2024 brought significant updates to server processors. AMD launched its 5th-Gen EPYC line (“Turin,” based on Zen 5) in October 2024, offering 8–192 cores on the established SP5 platform. Around the same time, Intel rolled out its 6th-Gen Xeon Scalable chips: Granite Rapids (up to 128 “P” performance cores) and Sierra Forest (up to 288 efficient “E” cores). Intel’s Sierra Forest was announced in June 2024, and Granite Rapids shipped in September of the same year. These new CPUs target demanding data-center workloads – from cloud and virtualization to AI inference. Below, we break down the latest performance, efficiency, and value comparisons between AMD’s and Intel’s server CPUs.
Early benchmark reports indicate that AMD’s Turin EPYC chips generally lead Intel’s new Xeons in multi-threaded throughput. For example, testing by Phoronix (cited in TechPowerUp) found that a dual-socket server with two 128-core AMD EPYC 9755 CPUs outperformed a comparable dual-socket Intel Xeon 6980P (Granite Rapids) system by about 40% on average across mixed workloads. In many cases, even a single 128‑core EPYC 9755 or 192‑core EPYC 9965 could keep up with Intel’s dual-socket system. Within AMD’s lineup, the new 128‑core 9755 shows roughly 1.55× the performance of its previous-gen 96‑core Genoa counterpart.
Intel’s latest Xeon “P” cores (Redwood Cove in Granite Rapids) offer solid gains over previous generations, but they generally trail AMD’s top-end chips in massive parallel workloads. In single-threaded or lightly threaded tasks, Intel’s new turbo frequencies and IPC improvements narrow the gap; however, AMD’s raw core count still gives EPYC the advantage in throughput. (Intel’s all-E-core Sierra Forest will boost thread counts in parallel workloads, but with each core weaker than a P-core.)
In synthetic and real-world tests, AMD’s claims are eye-catching. For instance, AMD notes that its 192-core EPYC 9965 can deliver up to 2.7 times the performance of the best Intel Xeon in specific AI benchmarks, and up to 4 times faster video-transcoding times. These figures are from AMD’s press materials, but they align with independent results: EPYC’s higher core and thread count consistently yields faster completion times on heavily parallel tasks. Overall, AMD holds a performance edge in raw multi-threaded workloads, while Intel’s refreshed cores aim to reclaim leadership in single-threaded speed.
Bacloud offers high-performance AMD EPYC dedicated servers with fast deployment — get your server up and running in just a few hours. Ideal for demanding workloads, virtualization, and modern cloud infrastructure. Reliable hardware, flexible configurations, and expert support included.
Get Epyc server now!
AMD’s Turin CPUs also excel in efficiency and cost-effectiveness. Because AMD uses a modern 5nm process and chiplet design, its new EPYCs deliver more performance per watt. In tests, the 192-core EPYC 9965 drew about 32% more power than the prior-gen Genoa (going from ~208W to 275W average), yet it produced 1.55 times the performance. In the same benchmarks, an Intel dual-Xeon setup (two 6980Ps) averaged ~322W (peaking at 547W) for less work, whereas the single-socket EPYC systems averaged only 275–324W. In other words, AMD’s chips are completing much more computation for each watt consumed.
Beyond raw watts, AMD’s pricing solidifies its value lead. AMD has aggressively priced its high-core CPUs; for example, the 192-core EPYC 9965 is priced at approximately $14,813, whereas Intel’s comparable 128-core Xeon 6980P is listed at $17,800. (Another analysis points out that Intel’s 128c Xeon is 37% more expensive than AMD’s 128c EPYC 9755.) Even Intel’s bigger 144-core Sierra Forest variant comes in cheaper ($11,350) than AMD’s 192c part – but it also delivers far less performance per socket. In practical terms, AMD chips offer far more throughput per dollar. Industry observers note that hyperscale cloud providers are voting with their budgets: it’s estimated that 50–60% of new hyperscale server deployments now use AMD processors, thanks to the performance-per-dollar advantage.
In short, AMD maintains leadership in raw efficiency. The Turin EPYCs combine world-class core counts with aggressive pricing, yielding superior performance per watt and performance per dollar. Intel will lean on Sierra Forest’s extreme core count (288 E-cores) and platform optimizations to close the gap, but as of late 2024, AMD holds the crown in both throughput and energy efficiency.
Intel still dominates overall x86 server CPU shipments, but AMD is making steady gains. According to Mercury Research, AMD’s share of the server CPU market rose to ~24.2% by Q3 2024 and about 27% by Q1 2025, with Intel at around 72–73%. In plain terms, roughly one in four new server CPUs now uses AMD silicon. This is a historic high – just a few years ago, AMD’s server share was in the single digits. (If we include desktop and laptop chips too, AMD’s total x86 share is lower, but the important takeaway is that AMD’s momentum is significant in servers.) Nonetheless, Intel remains the incumbent with the majority of installations worldwide.
Even with AMD’s technical edge, many enterprises continue to use Intel Xeon for legacy and ecosystem reasons. Intel’s Xeon line has decades of use in data centers, so its firmware, management tools and server BIOSes are extremely mature. Intel has extensive experience in the server market, manufacturing CPUs for servers, so its ecosystem is considered a benchmark for reliability. Many virtualization platforms, databases, and enterprise applications were initially developed and certified on Intel Xeon hardware. Independent software vendors typically optimize and test their software on Xeon servers first, meaning firms upgrading their existing infrastructure often face less integration risk if they stick with Intel.
Established reliability: Xeon firmware and drivers have been refined over many generations, giving firms confidence in uptime.
Software compatibility: Thousands of business-critical apps and OS images are “certified for Intel,” simplifying rollouts.
Vendor relationships: Major OEMs (Dell, HPE, Lenovo, etc.) and integrators have long built products around Xeon, so support and warranty programs are well-established.
Feature set: Xeon CPUs include mature technologies such as ECC memory support and Intel’s hardware accelerators (e.g., QuickAssist), which some deployments rely on.
In practice, this means that organizations with large existing Xeon fleets or conservative procurement policies may prioritize Intel due to its “known quantity.” Changing a data-center CPU is a big decision, and many IT teams prefer to stick with the vendor they’ve used for years, even if AMD offers better raw metrics.
In summary, AMD’s new EPYC “Turin” processors outperform Intel’s latest Xeons in many benchmarks, offering higher core counts, higher GHz per core frequency, and leading performance per watt and dollar. Intel’s Granite Rapids (P-core) and Sierra Forest (E-core) chips narrow the gap, particularly in scenarios that favor high clock speeds or extremely high thread counts. However, AMD currently holds the advantage in overall throughput and efficiency. Market share trends reflect this: AMD captures approximately 25–30% of the server CPU market share. Yet Intel’s entrenched position means Xeon remains the default CPU for many companies and licensed software.