Amazon AGI has narrowed to one bet: a flagship frontier model, due around re:Invent 2026

Asked:

What Amazon AGI team is focused on right now

A scan of current reporting and official AWS material through 2026-09-04, distilled to four active workstreams ranked by priority. The picture: engineering, compute and portfolio decisions are funneling into Frontier Model Research (FMR) — the internal initiative led by Pieter Abbeel — whose deliverable is a new flagship foundation model, possibly under the Nova name. FMR is a program, not a released model.

Ribbon width reflects reported priority rank (1 = highest), not budget figures. Grey ribbons mark efforts scaled back or closed. Hover a ribbon or node for detail.

What the funnel can’t say alone

The flagship model is a reported target for around AWS re:Invent 2026 — internal strategy, not an Amazon-confirmed launch date. eciks.org

The Indiana AGI SuperCluster consolidates data centers, deploys thousands of Trainium-powered servers, and expands storage for multimodal training, per internal planning documents. businessinsider.com

Amazon’s own confirmed language is narrower: large-model building stays one of its most important efforts while it sharpens focus on what matters most to customers — said as it closed its San Francisco AGI Lab. sfchronicle.com

Four Nova products survive the cut — Nova 2 Sonic, Nova 2 Lite, Nova Forge and Nova Act — while most of the broader Nova lineup is scaled back, per secondary reporting. cordcuttersnews.com

The four workstreams

Source rows

#	Workstream	Next milestone	Confidence	Source

Method: desk scan of current reporting and official AWS material available through 2026-09-04, reduced to a four-row priority view (one row per workstream, ranked by importance). Ribbon widths in the funnel encode priority rank only. Confidence labels distinguish Amazon-confirmed statements from reported internal strategy; reported plans are not presented as officially confirmed. Nothing was cut for space.

This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors. Nothing in this report is investment, legal, financial, or other professional advice. All trademarks and referenced content remain the property of their respective owners; no affiliation or endorsement is implied. To report an error, rights concern, or request removal: legal@keenable.ai.