ATD Calculator -- Supermicro SYS-522GA-NRT vs Dell PowerEdge XE9680

This interactive calculator requires JavaScript. Below is the reference data for a 100-GPU deployment over 5 years at $0.12/kWh with H200 GPUs ($30K each).

Metric	Supermicro SYS-522GA-NRT	Dell PowerEdge XE9680
Chassis needed (100 GPUs)	10	13
GPU density	2.0 GPUs/U	1.33 GPUs/U
Hardware CapEx	$3.45M	$3.12M
Power (5yr)	$4.20M	$5.13M
Support (5yr)	$0.09M	$0.42M
Total 5-year TCO	$33.5M	$36.4M
ATD advantage	Winner (+8% TCO savings)	Baseline

Enable JavaScript for the full interactive calculator. Read the full analysis.

Free Tool

ATD Calculator

Aggregate Throughput per Dollar for dense GPU AI server farms.

Compare Supermicro SYS-522GA-NRT (10 GPU, 5U) against Dell PowerEdge XE9680 (8 GPU, 6U) for your target deployment. Defaults reflect 2026 reference pricing and typical enterprise constraints.

Deployment Parameters

Target GPU countTotal GPUs needed across the full clusterGPU price (USD each)H200 ≈ $30K, B200 ≈ $50K (reference only)Power rate ($/kWh)US enterprise data center average: $0.10–$0.15/kWhCluster lifetime (years)Standard enterprise amortization: 5 yearsSustained TFLOPS per GPU (FP8)H100 ≈ 4000 TFLOPS FP8, B200 ≈ 10000 TFLOPS FP8Thermal sustain efficiency (0–1)Liquid cooled ≈ 0.97, high-end air ≈ 0.92, baseline air ≈ 0.85

Supermicro SYS-522GA-NRT

WINNER

Chassis needed: 10
Total rack space: 50U
GPU density: 2.00 GPUs/U
Hardware CapEx: $3.45M
Power (5yr): $420K
Support (5yr): $90K
Rack & networking: $85K
Total 5-yr TCO: $4.05M
ATD score: 93,932

Dell PowerEdge XE9680

Chassis needed: 13
Total rack space: 78U
GPU density: 1.33 GPUs/U
Hardware CapEx: $3.83M
Power (5yr): $512K
Support (5yr): $416K
Rack & networking: $111K
Total 5-yr TCO: $4.87M
ATD score: 81,084

Recommendation: Supermicro

TCO Savings

$828K

TCO %

17.0%

ATD Edge

+15.8%

Fewer Chassis

Over 5 years at 100 GPUs, the Supermicro SYS-522GA-NRT saves $828K (17.0%) versus the Dell PowerEdge XE9680. Supermicro needs 3 fewer chassis thanks to its 10-GPU density (versus Dell's 8-GPU limit), which compounds into lower power, networking, support, and rack overhead. The ATD (Aggregate Throughput per Dollar) advantage is +15.8%.

The ATD formula

Aggregate Throughput per Dollar is the economic model underneath this calculator. It normalizes raw compute against total amortized cost so dense configurations win structurally, not just on sticker price.

ATD = G × (TFLOPS_FP8 × Eff_thermal) / Σ(Node_cost + Rack_cost + Energy_5Y)

Where G = GPU density per chassis (10 Supermicro vs 8 Dell). The denominator costs scale per chassis, not per GPU — which is why higher density compounds into larger TCO and ATD advantages as cluster size grows.

Read the full head-to-head analysis Get your free ACRA Assessment

Reference defaults: Supermicro chassis $45K, Dell chassis $55K, Supermicro power 8.0 kW, Dell power 7.5 kW, Supermicro Gold OS4HR $1,800/chassis/yr, Dell ProSupport Plus $6,400/chassis/yr, rack & networking $8,500/chassis. Sourced from 2026 enterprise procurement reference pricing. Your actual costs may vary — substitute your quoted numbers in the inputs above.