Ericsson v D-Link
WiFi 802.11n SEP Pricing Prediction
Ericsson Inc. v D-Link Systems Inc. et al.
Data freeze: 1 September 2010 · September 2010 to December 2014, 4.2 years
An out-of-sample FRAND rate prediction. The platform predicted a WiFi per-unit rate of $0.094 to $0.300 before the case went to trial. The jury awarded $0.15 per device, landing at the 35th percentile of the predicted range. The Federal Circuit later vacated the damages, consistent with the platform flagging E.D. Texas as producing above-median FRAND rates.
5 of 7 dimensions predicted correctly. The FRAND rate fell within the predicted range on a case the models had never seen.
Our models predicted a WiFi FRAND rate of $0.094 to $0.300 per device, base $0.188. The jury awarded $0.15 per device, falling at the 35th percentile of the predicted range. The Federal Circuit later vacated the damages as improperly calculated, consistent with the platform flagging E.D. Texas as a venue that produces above-median FRAND rates.
Per-unit royalty, WiFi 802.11 SEPs | scale $0.00 to $0.35
Relative to baseline UK EWHC rate | higher factors indicate above-median damages
Analysis dimensions
FRAND Rate Model
The platform uses a scaling relationship calibrated from court-determined FRAND rates across cellular and WiFi standards. At model freeze, 1 September 2010, the model contained three WiFi-specific comparable rates, litigation outcomes for over 85,000 US patent cases, and venue-specific performance data for E.D. Texas. This case was never used in calibration or validation. It is a genuine out-of-sample test.
WiFi Comparable Rates
Three direct WiFi comparables existed at model freeze. Motorola v Microsoft, $0.036 per unit for ~100 families in W.D. Washington. In re Innovatio, $0.096 per unit for ~100 families in N.D. Illinois. CSIRO v Cisco, $0.830 per unit for 1 family in E.D. Texas. The wide variance between these three outcomes drives the width of the predicted range.
Jurisdiction Adjustment
Model data showed that E.D. Texas produces systematically higher patent damages relative to other circuits, with a jurisdiction factor of 1.2x. For FRAND-encumbered patents, the CAFC had been signalling stricter scrutiny of damages calculations. The platform identified a tension: E.D. Texas juries awarded higher rates, but appellate courts increasingly vacated SEP damages for improper instructions.
Device Volume Estimation
Estimating annual device shipments across six defendants required external revenue data not contained in the platform. The model used publicly available annual reports to approximate 24.6 million infringing devices per year across all defendants, 67 million over the 3-year infringement period. This is the weakest input parameter, as individual defendant allocations depend on product mix and pricing.
What We Cannot Predict
Two dimensions were beyond model coverage. Post-vacatur settlement dynamics, because no model exists for predicting how parties resolve after an appellate vacatur. Defendant counter-portfolio leverage, because no data existed on D-Link, Netgear, Dell, or Toshiba patent portfolios. These are specific, addressable data gaps.
Other case studies
Nokia v OPPO
SEP Enforcement Campaign Feasibility
6 / 6 predictions correctEricsson v TCL
SEP Licensing Enforcement, US Litigation
7 / 8 predictions correctMotorola v Microsoft
H.264 and WiFi SEP Licensing Enforcement
7 / 8 predictions correctRun it on your portfolio.
We generate this level of intelligence for your patents, your targets, your enforcement campaign.
Book a Call