Two Sides of the Same House: How Google Sells AI in 2026

Transcript

Full narration of the briefing. Click a section heading to jump the player to that moment.

Two sides of the same house 0:00

Google sells AI on two sides of the same house. On the seat side, Gemini is folded into Workspace editions and sold again as Gemini Enterprise, per user, whole population, with consumption overage on top. On the consumption side, Vertex AI, agents, Veo, Imagen, and the avatar and video features bill per token, per second, and per generation, inside a Google Cloud commitment. I am Tom, Claire is with me, and this is session one of the Google AI Negotiation Series, five briefings from the Redress research note on seats, consumption, and the commit that joins them.

One account team, two rooms 0:43

The structural fact first. The two sides are negotiated by different teams on the customer side, Workspace in one room and Cloud in another, and by one account team on Google's side. Google sizes the commit from both. The buyer that negotiates them separately funds the same floor twice.

This note is written for CIOs, security leaders, asset managers, digital workplace owners, and procurement with a Workspace renewal, a Gemini Enterprise proposal, a Vertex agent program, or a Google Cloud commitment on the table in the next twelve months. It is not tied to any single client situation, and it is written from the buyer's side.

The four takeaways 1:23

Four takeaways carry the whole note. Seat side: the standalone Gemini add ons are gone, the base price rose to absorb them, and Gemini Enterprise adds a second seat with an included quota that heavy users exhaust in days. The trap is paying for every user and leaving the overage open. Consumption side: introductory token rates on the newest models reset upward on January first, twenty twenty seven, media and avatar generation has no forecast, and Gemini Enterprise commit plans carry a symmetric downside.

The trap is committing to Google's forecast on a rate that is about to change.

One exercise, both sides 2:04

The response is one exercise across both sides. A bottom up forecast in Google's own meters. A pilot that decides who gets a seat. Ceilings on every line that cannot be forecast.

Unit rates fixed in the order form past the January reset. And a single ramped commit sized to your number rather than Google's. On the illustration the note works through, a mid sized enterprise with five thousand knowledge workers, Google's opening structure totals fifty million dollars over three years plus open lines; the buyer side structure lands at thirty seven point three million, every line capped.

How the series runs 2:40

Here is how the series runs. Session two covers the seat side: how Gemini is folded into Workspace editions, the Gemini Enterprise seat and its quota, Code Assist, Annual versus Flexible plans, and the true up that replaces the audit. Session three is the seat side response: reconcile, map editions to roles, pilot with a threshold, pool and cap the quota, protect the base rate, keep true down. Session four is the consumption side: tokens, agents, avatars, media, the commit plans, and the Google Cloud commitment.

Session five is the eight move response and the mistakes that cost the most.

The commit is the money 3:23

One number to hold before session two. Every token and seat discount is small next to the difference between a flat floor sized to Google's model and a ramped floor sized to yours. In the illustration that is a thirty six million dollar flat annual floor against twenty seven million ramped, with shortfall rollover. The commit is the money; everything else in this series is what makes the commit honest.

The full research note is free under this video, and the series lives at redresscompliance dot com slash google ai negotiation series. Next: the seat side.