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 →Yes—AWS increased prices for selected EC2 Capacity Blocks for ML by approximately 15% in early January 2026. The reported change affected H200-backed p5e.48xlarge and p5en.48xlarge offerings. AWS did not announce a blanket EC2 increase: ordinary On-Demand, Spot, Savings Plan and Reserved Instance prices were not shown to change in this event.
The practical distinction is crucial. A Capacity Block price is dynamic until you buy it, then locked for that reservation. A block purchased before the update keeps its original price; a future run that has only been planned or searched is exposed to the current offering price.
What changed in January 2026?
The Register reported on January 5, 2026, that AWS had raised published Capacity Block rates after a weekend pricing-page update. AWS attributed the adjustment to expected supply-and-demand patterns. The update was not accompanied by a prominent standalone announcement, although AWS had indicated that a pricing update was scheduled for January. “Quietly” describes the presentation of the change, not a separate AWS product category.
| Capacity Block instance | Previous hourly rate | New hourly rate | Approximate increase |
|---|---|---|---|
p5e.48xlarge, most regions |
$34.61 | $39.80 | 15.0% |
p5en.48xlarge, most regions |
$36.18 | $41.61 | 15.0% |
p5e.48xlarge, US West (N. California) |
$43.26 | $49.75 | 15.0% |
These are effective hourly reservation rates per EC2 instance, not an all-in workload bill. Operating-system licensing, storage, data transfer and monitoring can add to the total. The reported figures describe selected offerings; they do not establish that every Capacity Block, Region or accelerator became more expensive. (The Register, January 5, 2026)
#1 Best Overall
- [ Maximum AI Compute Power ] Dominate complex workloads with the ASUS ESC8000A-E13. This 4U rack server is a powerhouse engineered for mass-scale AI, machine learning, and deep training. Featuring support for dual AMD EPYC 9005/9004 processors and up to eight dual-slot GPUs, it delivers the raw computational muscle required to train LLMs and run complex simulations effortlessly. Accelerate your data science pipeline and transform raw data into actionable intelligence faster than ever.
- [ Advanced Thermal Efficiency ] High performance demands elite cooling. The ESC8000A-E13 features a cutting-edge aerodynamic design with independent CPU and GPU airflow tunnels. Equipped with redundant hot-swap fans and optimized for liquid cooling integrations, this 4U server ensures maximum uptime under heavy, sustained workloads. Keep your data center running cool, quiet, and highly efficient while preventing thermal throttling during mission-critical enterprise operations.
- [ Scale with Flexible Storage ] Future-proof your infrastructure with unmatched storage and expansion flexibility. This offers comprehensive front-panel drive bays supporting Gen5 NVMe, SAS, or SATA drives alongside multiple PCIe 5.0 slots. Designed as a high-density 4U server capable of housing eight dual-slot GPUs: NVD H200, RTX PRO 6000 Blackwell, RTX PRO 4500 Blackwell or AMD Instinct MI350P PCIe Card, each supporting up to 600 watts.
- [ Enterprise-Grade Reliability ] Minimize downtime and secure your ecosystem with server-grade redundancy. The ESC8000A-E13 is built for 24/7 continuous operation, boasting 2+2 redundant (3200W total) 80 PLUS Titanium power supplies and integrated ASUS ASMB11-iKVM for comprehensive out-of-band management. Ideal for cloud service providers, rendering farms, and large enterprise infrastructure, it combines robust physical hardware with smart remote monitoring to safeguard your digital assets.
- [Reliability Guaranteed] Shop with total peace of mind knowing that every new computer component we sell is backed by our EPC 3-year warranty. Whether you are investing in high-speed DDR5 RAM or a powerhouse GPU, we protect your build against defects and performance failures. We stand firmly behind the quality of our hardware, ensuring that your setup remains fast, stable, and secure for years to come.
What Capacity Blocks actually buy
Capacity Blocks for ML let you reserve accelerated EC2 capacity for a future start date and fixed duration. Instances are placed in EC2 UltraClusters designed for large-scale machine-learning workloads and high-performance networking. Typical uses include multi-node training, fine-tuning, short experiments, evaluation runs and scheduled inference that cannot risk waiting for GPUs to appear.
A block is useful when you need predictable access for days or weeks but do not want a longer commitment while hardware sits idle. It is not a promise that software will succeed or that a region will avoid every operational problem; it guarantees scheduled capacity subject to the service’s terms and the block’s forced end.
How AWS sets the price
Capacity Block pricing is explicitly supply-and-demand driven. You specify an instance type, instance count, future date range and duration; AWS returns eligible offerings and their prices, including the lowest-priced offering that meets the requested requirements when one is available. That makes the product closer to a date-sensitive reservation market than to a permanent regional EC2 price list.
Rank #2
- NVIDIA Volta GV100 Architecture — 4,608 CUDA Cores, 640 1st-Gen Tensor Cores delivering 14 TFLOPS FP32 and 112 TFLOPS deep learning performance for AI training, inference, HPC, and scientific computing workloads
- 32GB HBM2 ECC Memory — 900 GB/s Bandwidth — High-bandwidth memory on a 4096-bit bus with ECC error correction provides the memory capacity and throughput required for the largest AI models, simulations, and datasets
- PCIe 3.0 x16 Interface — 250W TDP — Standard PCIe Gen3 connectivity with passive cooling designed for enterprise rack server deployment in HPE ProLiant, Dell PowerEdge, and Supermicro platforms with adequate chassis airflow
- NVLink — Scale to 96GB Unified Memory — Connect two V100 GPUs via NVLink at 300 GB/s bi-directional bandwidth to scale GPU memory from 32GB to 96GB for larger AI training and HPC workloads
- Multi-Precision Computing — Supports FP64 (7 TFLOPS), FP32 (14 TFLOPS), FP16 (112 TFLOPS) and INT8 precision modes for flexible deployment across training, inference, and scientific simulation workloads
- Choose the accelerator instance type and number of instances.
- Enter the desired future start window and duration.
- Query the returned offerings for Region, start time, end time and total price.
- Check the usable runtime and termination window before committing.
- Purchase the selected offering; AWS charges the reservation fee up front.
- Launch instances against the reservation ID and monitor its state.
For discovery, AWS documents the EC2 API and CLI workflow. An abbreviated example is:
Free tools Windows power users keep installed
One-click scans. No signup required.
aws ec2 describe-capacity-block-offerings
--instance-type p5.48xlargeThe actual filters and required parameters vary by instance type and requested dates. Use the current AWS CLI reference rather than treating this abbreviated command as a complete purchase procedure. (AWS Builder Center)
Does the increase affect reservations you already bought?
| Your situation | Price treatment |
|---|---|
| Capacity Block purchased before a later update | Original reservation price remains locked. |
| Offering searched but not purchased | No price protection; a later result can be higher or lower. |
| Future start date, purchased after the update | Charged at the price shown when purchased. |
| Another training run booked later | Subject to the then-current, supply-sensitive offering. |
AWS’s pricing page says the reservation price does not change after purchase. Price protection therefore starts at the transaction, not when a team places a tentative plan on a calendar. (AWS Capacity Blocks pricing)
Rank #3
- AI-Optimized: Designed to support up to 4 GPUs, it is perfect for handling intensive AI and machine learning tasks, ensuring high performance and scalability for advanced computational needs.
- Intelligent Storage: Equipped with 8 hot-swappable 3.5" SATA/SAS drives (12Gbps), featuring SGPIO and temperature control, it ensures efficient data management and reliable storage performance.
- Robust Cooling: The system includes 3x 12038 hot-swap PWM fans and 2x 8038 rear fans, providing advanced thermal management to maintain optimal temperatures and ensure stable operation under heavy workloads.
- Rack-Ready: Comes with a pre-installed rail kit, allowing for quick and easy installation in standard 19-inch server racks, making it ideal for data center environments and enterprise setups.
- Versatile Connectivity: Offers USB 3.0 and the latest USB 3.2 Type-C ports, ensuring high-speed data transfer and compatibility with a wide range of peripherals and devices for enhanced connectivity options.
August 18, 2026: the catalog is already broader
The January figures should not be mistaken for today’s complete price list. The AWS public page viewed on August 18, 2026 showed additional P6e GB200, P6-B300, P6-B200, P5, P5e, P5en, P4d, Trn1 and Trn2 offerings. AWS says rates are updated regularly, so the following is a dated snapshot rather than a standing quote.
| Offering shown | Example location | Effective hourly rate per instance | Accelerators |
|---|---|---|---|
p6-b300.48xlarge |
US West (Oregon), US East (N. Virginia) | $93.60 | 8 × B300 |
p6-b300.48xlarge |
US GovCloud | $117 | 8 × B300 |
p6-b200.48xlarge |
Ohio, N. Virginia, Oregon | $82.368 | 8 × B200 |
p6-b200.48xlarge |
US GovCloud | $102.960 | 8 × B200 |
p5.48xlarge |
Atlanta Local Zone, N. Virginia, Ohio, Oregon, Northern California | $34.608 | 8 × H100 |
u-p6e-gb200x72 |
US East (Dallas) Local Zone | $761.904 per UltraServer | 72 × B200 |
u-p6e-gb200x36 |
US East (Dallas) Local Zone | $380.952 per UltraServer | 36 × B200 |
The same page showed Linux at $0 for the displayed accelerator examples, with additional hourly charges of $0.336 for Ubuntu Pro, $1.8432 for RHEL, $3.8592 for RHEL with HA and $0.125 for SLES. OS billing rules differ: Linux, Ubuntu Pro, RHEL and RHEL with HA use per-second billing, while SLES uses hourly billing with a one-hour minimum. (AWS pricing snapshot and billing rules)
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The costs beyond the headline GPU rate
- Operating system: Linux may be listed at zero, but commercial distributions add licensing charges.
- Storage and data: EBS volumes, snapshots, S3 storage and transfer can be material for large checkpoints and datasets.
- Operations: CloudWatch, orchestration, cluster-management infrastructure and engineering time are separate from the reservation.
- Unused capacity: You pay for the block even if a job finishes early, starts late or leaves instances idle.
- Discount assumptions: AWS says Savings Plans and Reserved Instance discounts do not apply to Capacity Blocks.
The upfront reservation and later usage-related charges can be reconciled in the Cost and Usage Report. AWS identifies the upfront fee with reservation/ReservationARN and usage-related anniversary billing with lineitem/ResourceID. (AWS billing documentation)
Rank #4
- Professional AI & Creator Workstation: AMD Radeon AI PRO R9700 GPU with 32GB GDDR6 is engineered for AI development, professional content creation, and compute-intensive workloads.
- Massive 32GB Memory Capacity: 32GB of GDDR6 memory on a 256-bit bus provides ample bandwidth for large AI models, 8K video editing, and complex 3D rendering.
- Advanced RDNA 4 with AI Accelerators: 64 Compute Units with 3rd Gen Ray Tracing and dedicated 2nd Gen AI Accelerators for groundbreaking AI performance and visual computing.
- Professional Blower Cooling: Efficient single blower design exhausts heat directly out of the chassis, ideal for multi-GPU workstation and server configurations.
- Enterprise-Grade Thermal Solution: Vapor chamber heatsink with industrial Honeywell PTM7950 thermal interface material ensures reliable cooling under sustained professional loads.
Forced termination changes the usable runtime
For ordinary instance-type Capacity Blocks, instances can be used until 30 minutes before the block ends. UltraServer Capacity Blocks have a 60-minute cleanup window. AWS emits an EventBridge event 10 minutes before termination begins, and running instances are terminated as the block expires. (AWS Capacity Block operation)
- Checkpoint model state before the 30- or 60-minute cutoff.
- Upload checkpoints, logs and metrics before the final window.
- Subscribe automation to the EventBridge warning.
- Reserve staging and teardown time rather than scheduling the final epoch at the boundary.
- Test checkpoint restoration before the paid run.
Payment, cancellation and capacity limits
Payment state matters
AWS generally bills within five minutes to 12 hours after purchase. While processing, a reservation can remain payment-pending. If payment cannot be processed at least five minutes before the block starts, or within 12 hours—whichever comes first—the block is released and changes to payment-failed. Confirm payment capacity, permissions and billing alerts well before the training date.
Check limits before designing the run
- One Capacity Block can contain up to 64 instances.
- A customer can have up to 256 instances across Capacity Blocks.
- A block can be scheduled up to eight weeks in the future.
- A 64-instance block is not supported for every instance type in every Region.
(AWS Capacity Block limits and availability)
Do not assume you can unwind a purchase
AWS’s launch documentation said Capacity Block reservations could not be modified or canceled after purchase. Current pricing pages emphasize price locking and payment but do not restate every cancellation term. Confirm the service-specific terms in the console and with AWS before treating a reservation as refundable or changeable. (AWS launch announcement)
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Best Value
Should you still use a Capacity Block?
Usually a good fit when
- A deadline makes GPU availability more valuable than hourly flexibility.
- You need several colocated GPUs for a defined multi-day or multi-week run.
- Spot interruptions would invalidate the schedule.
- The pipeline has tested checkpointing and a fixed start and end time.
Usually a poor fit when
- Work timing, duration or instance count is uncertain.
- Utilization is low or jobs often finish early.
- You need cancellation, resizing or regional mobility.
- The workload tolerates interruption or can use smaller accelerators.
- The requested date, Region or count is not reliably offered.
The premium is for scheduled, colocated capacity—not automatically the lowest GPU-hour price. Compare the cost of a missed deadline, interruption recovery and engineering work with the reservation premium.
How alternatives differ
| Option | Main strength | Main limitation |
|---|---|---|
| Capacity Block | Scheduled accelerator capacity for a defined window | Upfront, dynamic pricing and forced end |
| On-Demand | Start and stop flexibility | Capacity may be unavailable when needed |
| Spot | Potentially lower price | Interruption and reclaim risk |
| Savings Plan | Discount for sustained eligible usage | Does not discount Capacity Block fees |
| Reserved Instance | Predictable pricing for eligible EC2 models | Not a substitute for scheduled Capacity Block capacity |
| Trainium or newer accelerators | May provide a different performance-cost profile | Requires software and availability validation |
| Third-party GPU cloud | Potentially different terms and regions | Networking, security, egress and availability differ |
| On-premises cluster | Long-term control | Capital cost, procurement time and operational burden |
Do not compare nominal rates without matching GPU generation, GPU count, interconnect, Region, duration, interruption behavior, OS and ancillary charges. AWS’s earlier GPU reductions of up to 45% applied to On-Demand and Savings Plans, not directly to the Capacity Block product discussed here. (The Register)
A practical procurement checklist
- Query several date ranges and Regions instead of assuming one permanent rate.
- Match the instance family to model memory, networking and software requirements.
- Add startup, data staging, checkpointing and teardown time to the requested duration.
- Calculate the total reservation fee plus OS, EBS, transfer and monitoring costs.
- Confirm quotas, payment capacity and account permissions.
- Purchase only after validating checkpoint recovery and workload readiness.
- Record the reservation ID, start and end times, and cost-allocation tags.
- Monitor payment state and termination events programmatically.
- Ask the AWS account team how public Capacity Block prices interact with your negotiated contract; do not assume an enterprise discount protects the old absolute price.
- Keep an alternative execution path for work whose deadline can move.
What this increase does—and does not—prove
The evidence supports a narrow conclusion: selected public Capacity Block rates rose, and AWS describes the product as responsive to supply and demand. High-end accelerator scarcity is a plausible market context, but AWS did not publish a detailed cost breakdown tying the January change to a particular chip, memory or supplier expense.
The event does not prove a broad increase across EC2, nor does it establish that storage, networking or all GPU products will follow. It also does not determine how every negotiated enterprise agreement treats the new public baseline.
Recommended Free Tools
The Bottom Line
Bottom line: the January 2026 increase matters most for teams booking large, time-sensitive H200 runs. Existing Capacity Blocks are price-protected, but future offerings remain dynamic. Check the live offering, total bill, payment state and termination window before committing; Capacity Blocks buy scheduling certainty, not a permanently fixed GPU price.
Quick Recap
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.
