The forecast is a multiplication, and it is worth doing by hand. Barclays has 2,485 billion ChatGPT queries in 2030 at $0.041 of revenue each [11][13], which comes to $101.9bn, the headline number [14]. Volume carries part of that: queries rise about 2.5 times from 1,002 billion in 2026, on an advertisable user base going from 0.7 billion to 1.7 billion [11][10]. Price per query carries the rest, rising roughly twentyfold from $0.002 [15]. A model that turns $2.4bn into $102bn, a 42.5-fold increase [21], is therefore mostly a statement about monetisation rather than about usage.
Barclays is explicit about the endpoint. The 2030 figure of $0.041 would be 96% of Google's current search revenue per query [13], a denominator the bank builds from 2025 revenue and 5.25 trillion queries [12], which works out near $0.043 on the $224bn line [16]. What the model assumes, then, is that conversational placements come to monetise commercial intent at near-parity with a surface built on purchase intent signals refined over 25 years [22]. Whether that holds is not something the record settles.
A skeptic would say a four-year sell-side model is a spreadsheet, and that a constant 4.0 queries per user per day, carried unchanged across the whole projection from a separate Barclays note on ChatGPT usage [11], is an assumption dressed as a finding. Fair. Which is why the useful output is the ordering rather than the total.
Set against Google's structure, $102bn is about 45% of the $224bn search line [17] and more than YouTube advertising and Google Network combined, which came to roughly $71bn in 2025 [8][18]. Alphabet's 2025 advertising total was $294.69bn according to Reuters [4], so the target is a little over a third of everything [26]. The distribution matters more than the fraction: a forecast aimed at one of three structurally different lines [5] is a claim on roughly half of one business, not a third of three.
The near-term figure carries its own slack. At 1,002 billion queries and $0.002 each, 2026 arithmetic yields about $2.0bn, under the $2.4bn Barclays prints for that year [19]. OpenAI's own guidance, $2.5bn this year and $100bn by 2030 as Reuters reported [3], lands within 2% of the sell-side total at the far end [20], so a planner holding both is holding one estimate twice. For this quarter that supports a small experiment against a line whose reported economics remain a rounding error next to $224bn [7]; for the back half of the decade it leaves a pricing question on which nobody yet has evidence.
