FTC pins $1 billion of excess profit on six years of Amazon's Nessie pricing algorithm

Divide the FTC's number by the years and the per-unit edge gets small. More than $1 billion, or as much as $1.4 billion on some estimates in the case, covers 2014 through 2019 ^[3], which is roughly $167 million to $233 million a year ^[1]. April 2018 supplies a denominator: Nessie priced more than 8 million items and shoppers looked at them over 400 million times ^[6], about 50 views per item ^[2]. If that month was typical, the year holds 4.8 billion pageviews, and the alleged excess profit works out to between three and a half and five cents a view ^[3].

What the tool needed was not an agreement. On the FTC's account it looked for products where competitors were likely to match an Amazon increase, raised the price, and held it once they did ^[2], and many of those competitors were running automated repricing software of their own ^[17].

Court documents say the algorithm was switched on and off at least eight times between 2015 and 2019 ^[7], about one flip every seven or eight months ^[4]. The FTC alleges the pauses came during periods of heightened regulatory scrutiny, with the system restarted when attention moved on ^[8]. Amazon paused it in 2019, as antitrust interest from Congress and federal regulators was increasing ^[9], and internal documents reviewed in the case suggest it was reactivated in experiments as late as 2022 ^[10]. Amazon says the tool was discontinued years ago and never worked the way the government describes ^[4].

The Nessie claims sit inside the complaint the FTC filed in September 2023 with 17 state attorneys general, alleging maintained monopoly power in online retail ^[14], along with penalties for sellers who offer lower prices on other platforms and search results degraded in favour of Amazon's own products ^[15]. Certain state consumer-protection claims tied to Nessie were dismissed in March 2025, and the core federal antitrust claims survived ^[16]. Three years pass between the filing and the October 2026 trial date ^[5].

The estimate the agency has to defend has a $400 million spread inside it, 40 percent of the low end ^[7]. Amazon's position is that Nessie was limited in function, was meant to prevent unsustainable price drops, and was shut off because it did not work effectively ^[11]. A per-item accounting showing increases reversed when rivals declined to follow would take the damages theory apart. Pulling the other way, FTC filings reference internal documents that characterised Nessie's performance as "an incredible success" ^[12], and records suggest Amazon leadership explored reviving the program after it was officially shelved ^[13]. In my view the eight-flip toggle log will matter more at trial than the billion-dollar figure, because a switch has a date and a damages model has a range.