SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Qualcomm Moderated Thread - please read rules before posting
QCOM 178.51-1.4%3:59 PM EDT

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: QCOM_HYPE_TRAIN who wrote (196266)10/28/2025 1:46:09 PM
From: Jim Mullens8 Recommendations

Recommended By
Dr. John
John Hayman
kech
kozom
manning18

and 3 more members

  Read Replies (1) of 196305
 
QHT, re: Inference Data Center Spare- Nvda BOM and Energy (TCO) vs QCOM solution-

This is what is so exciting about entering into this space. Nvidia racks are so wildly expensive that when QC come in with mostly competitive inference racks, but at a significant discount its a win-win all around.

DCs can get inference capabilities, QC gets revenue that was never forecasted -- so QC can undercut.


>>>>>>>>>>>>>>>
Per Copilot- thought this was an interesting comparison.

Can you verify the reasonableness ???

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

Total Cost of Ownership (TCO) & Energy Cost Comparison: Qualcomm vs. Nvidia vs. AMD

Qualcomm’s AI250 racks deliver up to 60% lower energy cost and 45–65% lower TCO per rack compared to Nvidia Rubin and AMD MI300X, driven by LPDDR5X memory, near-memory compute, and 160 kW rack power vs. 500–600+ kW for competitors.



Total Cost of Ownership (TCO) — 3-Year Estimate per Rack

Vendor

Platform

CapEx (USD)

Energy Cost (3 yrs)

TCO (3 yrs)

vs. QCOM AI250

Qualcomm

AI250

$332,000

~$168,000 (160kW @ $0.04/kWh)

$500,000

— (baseline)

Qualcomm

AI200

$211,000

~$168,000

$379,000

24% lower TCO

Nvidia

Rubin NVL576

$741,000

~$630,000 (600kW)

$1.37M

63% higher

Nvidia

GH200 Grace Hopper

$700,000

~$420,000 (400kW)

$1.12M

55% higher

AMD

MI300X

$303,000

~$525,000 (500kW)

$828,000

39% higher

AMD

MI300A

$330,000

~$570,000 (540kW)

$900,000

45% higher

Assumes 24/7 operation, $0.04/kWh industrial rate, 90% utilization, and 3-year amortization.



Energy Cost Breakdown (3-Year, per Rack)

Platform

Power Draw

Annual Energy (kWh)

3-Year Energy Cost

QCOM AI250

160 kW

~1.4M kWh

~$168,000

Nvidia Rubin

600 kW

~5.3M kWh

~$630,000

AMD MI300X

500 kW

~4.4M kWh

~$525,000


Key Efficiency Drivers for Qualcomm

  • LPDDR5X vs. HBM3/3e: Lower voltage, lower cooling load
  • Near-memory compute: Reduces data movement energy
  • Direct Liquid Cooling: Cuts HVAC overhead by up to 40%
  • ARM-based SoC: Lower idle and orchestration power
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext