Scuttlebutt

Scuttlebutt

AI agents see no ads, so the web's ad-funded model has to be replaced, says Parallel CEO

20 predictions: why web search prices must fall by an order of magnitude, what Amazon risks by blocking agents, and how 5–20% of AI inference spending could make Parallel a $100B company

Scuttlebutt's avatar
Scuttlebutt
Sep 27, 2026
∙ Paid

The Twenty Minute VC (20VC) · Sep 26, 2026 · 57m · 36m saved

With Parag Agrawal, Founder and CEO of Parallel, previously CEO of Twitter

AI agents see no ads, so the web's ad-funded model has to be replaced, says Parallel CEO

Parallel charges $1 for every 1,000 web searches that its customers' AI agents run. Parag Agrawal said almost every rival charges $7, $10 or $14 for the same volume.

Today's web search prices were set by what Google earns from ads shown to people. Agrawal said agents will search the web 1,000 times more than people do and will never look at an ad, so both the price of search and the way publishers get paid have to change.

"Agents show up, no one sees ads, you make no money."

Parag Agrawal, founder and CEO of Parallel, on The Twenty Minute VC. He ran Twitter as chief executive before Elon Musk bought it, and his company, which sells web search to AI agents, counts Vinod Khosla on its board.

The full interview is covered here so you can skip it. 57 minutes of audio, 21 minutes of reading.

Here are the 20 predictions that matter.


This Scuttlebutt Summary is available in full to premium subscribers. If you're not one already, please consider subscribing to keep reading.


Key Takeaways

  • Agents will use the web 1,000 times more than people, and Agrawal said no technology survives that jump in scale unchanged

  • Coding agents call web search in about 5% of prompts, while law, insurance, sales and science lean on it heavily

  • Parallel pays publishers each time an agent benefits from their content, to replace the ad revenue agents do not generate

  • Parallel charges $1 per 1,000 searches against $7–$14 at rivals, and Agrawal thinks another 10x cut is possible

  • Agents should spend less on search than on the model reading the results, but some setups now spend 80–90% on search

    • He traced today's prices back to Google's ad economics

  • He puts web search at 5–20% of AI inference spending, the basis for his $100B case for Parallel

  • Amazon's refusal to let agents in is only right if its own agent succeeds

  • Most people are not ready to give an agent their bank login, whatever the technology can do

  • He admires Elon Musk's urgency and "unreasonable expectations of people", despite many disagreements

1. A Google for Agents

The host opened by saying he had called Vinod, Andrew Reed and Todd Jackson before the interview. He asked Agrawal to describe Parallel in 60 seconds.

Parallel sells web search built for AI agents rather than people

"Parallel is the Google for agents." — Parag Agrawal

Agents need to search the web to do almost anything for their users, whether they are personal assistants or work tools, in the same way people use Google during the day.

"Turns out agents are different from humans, and the way you build web search for agents is different — so Parallel is about building the technology for agents to search the web, and then the business models to make that sustainable." — Parag Agrawal

The founding idea was a 1,000-fold jump in web use

Agrawal said the company began with one written statement: agents will use the web a thousand times more than humans, so new technology and new business models are both needed.

"A thousand-x gives you a sense of scale — it changes how you think about building the tech underneath, because no tech built for a certain scale survives three orders of magnitude." — Parag Agrawal

2. Search Built for Machines

User's avatar

Continue reading this post for free, courtesy of Scuttlebutt.

Or purchase a paid subscription.
© 2026 The Golden Age LLC · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture