Estimation question

Estimate the annual revenue Anthropic makes from Claude Code.

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Tests structured estimation: building a revenue estimate from user segments and pricing tiers for a specific product line.

Segment Claude Code users into tiers: individual developers on a subscription, and enterprise teams on usage based or seat based contracts.
Estimate individual users: assume 2 million active developers, with 20 percent paying an average of 20 dollars a month, giving 400,000 paying users.
Calculate individual revenue: 400,000 users times 20 dollars times 12 months is 96 million dollars a year from individual subscriptions.
Estimate enterprise revenue separately: assume 5,000 enterprise teams paying an average of 50,000 dollars a year in usage based API spend, giving 250 million dollars.
Sum the two: roughly 346 million dollars a year, and round to state an estimate in the range of 300 to 400 million dollars annually.
State clearly that user counts, conversion rate, and average spend are assumptions you would sanity check against any public signal available.

Splits the estimate into individual and enterprise revenue, since the two have very different price points and should not be blended into one average.
Shows the full calculation chain for each segment rather than jumping to a final number.
Labels every input, 2 million developers, 20 percent conversion, as an assumption, and gives a reasoned range rather than false precision.
Sanity checks the result against what is publicly known about Anthropic's overall revenue scale, noting Claude Code would be a meaningful but not dominant slice.

Blending individual and enterprise users into one average price, which distorts the estimate significantly.
Presenting a single precise number without a range, implying false confidence.
Not stating which inputs are assumptions versus known facts.

How would you adjust this if enterprise usage based spend is much lumpier than assumed?
What signal would tell you this estimate is too high or too low?
How would you estimate the growth rate of this revenue line next year?