Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
AMD’s ZT Systems acquisition, its Rapt collaboration and Oracle’s EPYC-powered cloud instances address different parts of the AI infrastructure stack. Together they show AMD’s effort to compete beyond individual chips—but they do not, by themselves, establish Nvidia-level software parity, guaranteed customer savings or broad availability of AMD GPU capacity in Oracle Cloud.
Three moves, three different jobs
The three announcements often grouped together in 2025 were not one product launch. They addressed systems engineering, GPU workload operations and cloud computing:
| Move | What it covers | Customer problem it may address |
|---|---|---|
| ZT Systems acquisition | Rack-scale systems design and customer enablement | Integrating processors, accelerators, networking and other components into deployable data-center systems |
| Rapt collaboration | GPU workload automation and optimization | Scheduling and managing training or inference workloads across GPU environments |
| Oracle OCI Compute E6 | Cloud instances powered by AMD EPYC CPUs | Accessing CPU compute without buying and operating servers |
The strategic connection is that AMD wants its hardware to be easier to design into systems, operate and access through cloud services. The announcements are not evidence that every layer is already integrated into a single AMD offering.
Recommended Free Tools
ZT Systems: from acquisition announcement to completed deal
AMD announced its plan to acquire ZT Systems on August 19, 2024, and completed the acquisition on March 31, 2025. ZT designed, manufactured, integrated and deployed data-center infrastructure, particularly for large cloud and hyperscale customers. Its experience was relevant to AMD’s effort to sell into large AI deployments, where the challenge is not simply choosing a GPU.
#1 Best Overall
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- 0dB technology lets you enjoy light gaming in relative silence
- Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles
- Dual ball fan bearings last up to twice as long as sleeve bearing designs
A production AI cluster depends on the interaction of processors and accelerators with networking, memory, storage, power delivery, cooling, firmware, validation and serviceability. Rack-level design can affect how much compute a customer can deploy, how reliably it runs and how difficult it is to maintain. AMD’s stated rationale was to combine ZT’s systems expertise with AMD EPYC CPUs, Instinct GPUs, networking products and ROCm software. The acquisition was intended to help customers move from component selection to complete deployments more quickly.
The deal’s value needs a date and a definition. AMD announced a headline transaction value of $4.9 billion, including up to $400 million in contingent consideration. AMD’s later filing reported $4.4 billion in purchase consideration for the completed acquisition. Those are different measures—the announced value and the accounting consideration at closing—not necessarily contradictory prices. See AMD’s acquisition announcement, its completion announcement and the later filing.
AMD said ZT’s design teams would join its Data Center Solutions organization. AMD executive Forrest Norrod was to lead that organization, with ZT founder Doug Huang in a platform-engineering leadership role. Those details reflect the acquisition’s emphasis on design and customer enablement, not simply on owning a factory.
Why AMD later sold ZT’s manufacturing operation
AMD announced on May 19, 2025, that it would sell ZT Systems’ U.S.-based data-center infrastructure manufacturing business to Sanmina for $3 billion in cash and stock, including a contingent payment of up to $450 million. AMD completed the divestiture on October 27, 2025. It retained ZT’s design and customer-enablement teams, while Sanmina became a preferred manufacturing partner for new product introductions. Frank Zhang took a role leading the divested manufacturing business.
This was a consequential distinction: AMD did not reverse its interest in systems engineering. Instead, the transaction separated the capabilities AMD identified as strategically close to its silicon and platform work from production manufacturing. The arrangement lets AMD retain design and customer-facing expertise while relying on a manufacturing partner to build systems. It does not mean AMD itself manufactures every system built around its products. Details are in AMD’s divestiture agreement and completion announcement.
Rank #2
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- Phase-change GPU thermal pad helps ensure optimal heat transfer, lowering GPU temperatures for enhanced performance and reliability
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- Dual-ball fan bearings last up to twice as long as standard conventional sleeve bearings designs
- 0dB technology lets you enjoy light gaming in relative silence
What Rapt contributes—and what remains unproven
The reported AMD-Rapt collaboration focuses on software for automating and optimizing workloads on AMD Instinct GPUs. Described areas include managing training and inference workloads, scheduling, resource allocation and memory utilization. The coverage also says the platform is intended to operate across on-premises and multicloud environments.
That positions Rapt at the operations layer: deciding how workloads use available GPU resources and coordinating their execution. It is not a substitute for ROCm, AMD’s software ecosystem for developing and running workloads on its accelerators, nor does it replace GPU hardware, compilers or model libraries. In practice, infrastructure buyers need both compatible software and operational tools that fit their environment.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The available reporting does not establish independent benchmark results, quantified utilization improvements, supported Instinct models, detailed architecture, pricing or customer case studies. Claims that the software improves utilization or reduces total cost should therefore be treated as objectives or vendor assertions, not guaranteed outcomes. The phrase “out of the box,” used in descriptions of the collaboration, should not be taken to mean every GPU, scheduler, cloud or configuration will work without validation. The reported scope is summarized by Network World.
Oracle’s original announcement was about EPYC CPU instances
On March 31, 2025, Oracle announced OCI Compute E6 Standard shapes powered by 5th Gen AMD EPYC processors. The offering included bare-metal and virtual-machine options for general-purpose and compute-intensive workloads. Regions named at the announcement included Ashburn, Phoenix, Chicago, Frankfurt and London; that launch-era list is not a guarantee of current availability in every region. Consult AMD’s announcement and Oracle’s live service information for current offerings and capacity.
AMD said E6 could provide up to twice the performance of the previous E5 generation at the same price. That is a vendor claim tied to comparisons described in the announcement, not a universal result for every application or customer bill. Actual performance depends on workload and configuration, while cloud cost also depends on region, instance choice, discounts, storage, networking and other charges.
Rank #3
- System Compatibility Note: This 2‑slot card measures 249 mm (L) x 132 mm (W) x 41 mm (H) and requires a single 8‑pin power connector. Please verify available chassis clearance and ensure your power supply is rated for a recommended 550W before purchase.
- Dedicated Support: Please contact us directly through Amazon for any product questions or assistance you may require.
- Next‑Gen AMD RDNA 4 Architecture: Powered by the AMD Radeon RX 9060 XT GPU with 32 Compute Units featuring 3rd Gen Ray Tracing and 2nd Gen AI Accelerators, delivering exceptional 1440p gaming and AI‑enhanced performance.
- Blazing‑Fast Engine Clock: Delivers a boost clock of up to 3290 MHz and a game clock of 2700 MHz out of the box, providing the raw power for smooth, high‑framerate gameplay.
- 16GB GDDR6 Memory on 128‑Bit Bus: Equipped with 16GB of high‑speed GDDR6 memory running at 20 Gbps, offering ample capacity and bandwidth for modern game textures and creative applications.
Most importantly, an EPYC cloud instance is not an Instinct GPU instance. EPYC is a CPU platform suited to general cloud compute, databases, analytics, virtualization and CPU-heavy supporting tasks. AI training and many inference workloads may require accelerators, and a CPU instance listing does not establish that AMD Instinct GPUs are available at the required scale. Buyers should check the exact accelerator, region, quota, networking, framework support and reservation terms.
What changed after the 2025 roundup
- March 31, 2025: AMD completed the ZT Systems acquisition and Oracle announced EPYC-powered OCI Compute E6 Standard shapes.
- May 19, 2025: AMD announced its agreement to divest ZT’s U.S. manufacturing business to Sanmina.
- October 2025: AMD completed that divestiture. Separately, Oracle and AMD announced plans for an OCI AI supercluster using an initial 50,000 AMD Instinct MI450 GPUs, with deployment expected to begin in calendar Q3 2026.
The MI450 plan is a later development, distinct from the E6 CPU-instance announcement. It was announced as a future deployment, not as proof that the planned cluster was already operational. See the Oracle-AMD announcement for the stated plan and timing.
Does this make AMD a complete Nvidia alternative?
It strengthens AMD’s case as an alternative across more of the infrastructure stack. AMD supplies EPYC CPUs and Instinct GPUs; ZT added systems-design and customer-enablement expertise; ROCm addresses part of the software ecosystem; Rapt is aimed at GPU workload operations; and cloud partnerships offer routes to consume compute without buying hardware.
But the announcements do not prove parity with Nvidia’s CUDA ecosystem, networking portfolio, software maturity, system availability or customer adoption. Nor does support for ROCm mean every CUDA-dependent application will run unchanged. The defensible conclusion is that AMD is building a broader option for selected deployments, with potential benefits in choice and supplier diversity—not that it has matched every Nvidia product or removed migration work.
Quick Recap
What infrastructure buyers should verify
- Match the service to the workload. Determine whether the application needs CPU compute, GPU acceleration or both. Do not infer GPU capacity from an EPYC instance announcement.
- Validate software on the exact configuration. Check framework, library, container, driver and ROCm support for the specific GPU and deployment environment. Test representative models rather than relying on peak theoretical throughput.
- Confirm actual capacity and support. Verify instance type, region, quota, available capacity, reservation terms and support arrangements. A service announcement or regional launch list does not promise unlimited production capacity.
- Model total cost, not just chip or instance price. Include compute, storage, networking and egress, orchestration, support, engineering effort and utilization. Migration or optimization work can offset savings elsewhere.
- Check Rapt’s fit before assuming gains. Ask which GPU generations, schedulers, Kubernetes environments and clouds are supported, and confirm licensing and deployment terms. Compare measured results on your own workload; no quantified independent gain is established by the cited coverage.
- Account for operational dependencies. An AMD-based deployment may involve AMD, a cloud provider, systems integrators and software vendors. Assess support boundaries, incident response and the effort required to maintain the complete stack.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
