SEANBURNS
D / 05DisruptedPUBLICATION: July 2024EXPANDED: September 7, 2026

Generative AI · Content moat · Customer switching

Chegg vs. ChatGPT

A proprietary answer library loses scarcity when a general model can generate an answer at the moment of demand.

CENTRAL QUESTION

Was Chegg’s asset the content, the workflow, the brand—or simply the friction that AI removed?

01 / THESIS

The argument

Chegg’s subscription economics rested on turning a large answer corpus and student demand into a paid, repeatable workflow. Generative AI changed the competitive boundary: students could ask novel questions conversationally, receive explanations immediately, and switch tools with little cost.

In April 2023 Chegg announced an OpenAI partnership and a pivot toward AI. That response was strategically necessary, but it also revealed the dilemma. If model capability becomes broadly available, embedding the same capability may defend product relevance without restoring the scarcity that supported pricing power.

AI pivot announcedApril 2023
Switching costLow
Primary analytical variableContent scarcity

02 / CHRONOLOGY

The sequence

  1. PRE-2022

    Chegg scales a subscription product around expert-created solutions, search, and a recognizable student workflow.

  2. LATE 2022

    Consumer generative-AI tools make open-ended explanation and answer generation widely accessible.

  3. MAY 2023

    Chegg reports the emerging effect on customer growth and details an AI-centered product response.

  4. 2023—2024

    The company invests in AI services and proprietary models while confronting weaker assumptions about its prior moat.

03 / MECHANISM

How the failure compounds

01

Marginal-cost collapse

Generated explanations reduce the scarcity value of a pre-written answer archive.

02

Interface migration

Search-and-retrieve workflows can lose to conversational systems that infer intent and generate tailored responses.

03

Supplier commoditization

Using the same foundational capability as competitors can improve the product while shifting bargaining power to the model layer.

04 / JUDGMENT

What survives the case

The moat question must move from “How much content exists?” to “What proprietary signal improves outcomes with use?” Defensible assets might include verified pedagogy, longitudinal student data, accreditation, workflow integration, or trusted assessment—not answer volume alone.

For investors, the earliest signal was not revenue decline but a reduction in the cost and friction of the customer’s alternative. When substitution becomes nearly instantaneous, historical retention and acquisition economics deserve a structural rather than cyclical haircut.

EVIDENTIARY LIMIT

The long-run competitive outcome of AI-enabled education remains unsettled. This brief analyzes the change in industry structure, not the eventual winner.

05 / SOURCE DOCKET

Follow the evidence.

VERSION 1.0 · EXPANDED SEPTEMBER 7, 2026 · MATERIAL CORRECTIONS WILL BE RECORDED ON THIS PAGE.