Escape Meridian

Bridgewater Associates · Nov 26, 2025

Greg Jensen and Jas Sekhon argue in Bridgewater Associates that Gemini 3 intensifies AI’s resource grab

Markdown · JSON · Original source

On November 26, 2025, Greg Jensen and Jas Sekhon argue that Google’s release of Gemini 3 signals an intensifying AI “resource grab.” They say the model’s performance suggests scaling can continue, after challenges in training pipelines raised doubts about that approach. In their view, Gemini 3 puts pressure on other labs to invest in compute, chips and power, with major implications for capital markets and the economy. The authors also describe Google’s AI chips and hardware position as a competitive challenge to , and say that pressure helps explain Nvidia’s investments elsewhere in the AI ecosystem. They forecast that wider adoption could follow when businesses outside AI face pressure from AI-using competitors. But current models still require technical skills, subject matter expertise and meaningful work to produce large productivity gains; the authors say Gemini 3 makes that challenge more surmountable. They conclude that markets underappreciate the capital needs and economic boost over the next two years. Whether capability gains translate into productivity improvements and wider adoption remains unknown.

Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.

Bridgewater AssociatesArticle · bridgewater.comOpen original source

Original source

Bridgewater Associates · Article

Source details

Original source date: Nov 26, 2025 (2025-11-26).

Date basis: Date stated on the publisher’s page.

Added to Escape Meridian: 2026-11-18T12:18:17.956Z. First observed: 2026-11-18T01:44:02.424Z. Retrieved: 2026-11-18T01:47:58.459Z.

Referenced by

Recorded citations in public readings. A reference alone does not establish agreement or influence.

No public references recorded here yet.

Reference data (JSON)

Limits

Text as retrieved on November 18, 2026; originally published November 26, 2025. The authors’ assessment of Gemini 3’s performance draws on external and internal testing; the cited passage does not provide test results or methods. The authors describe their estimate of Gemini 3’s compute as a rough approximation.

Open question

What evidence would show that Gemini 3’s capability gains lead to productivity improvements and wider adoption beyond the AI industry?

Source-stated

Greg Jensen and Jas Sekhon say Gemini 3’s performance indicates that pre-training scaling can continue after concerns that training challenges might slow it.

But Google cracked the challenge with Gemini 3. It is the first model released with significantly larger, effectively used pre-training compute than GPT-4o, and it showed sizable capability improvements.

Source-stated

The authors argue that Gemini 3 will add pressure on the AI ecosystem to invest in compute, chips and power, with implications for capital markets and the economy.

The path to getting better models requires more compute, more chips, and more power.

Source-stated

Jensen and Sekhon argue Google’s AI chips and hardware position challenge Nvidia and describe Nvidia’s investments elsewhere in the ecosystem as a response to competitive pressure.

This competitive pressure from Google has been a key motivating factor behind Nvidia’s extensive downstream investments in the AI ecosystem—most notably OpenAI

Source-stated

The authors say wider AI adoption could follow when businesses outside the AI ecosystem face pressure from an AI-using competitor, while noting that current models require substantial work and expertise to use for large productivity gains.

Current frontier AI models are still not easy to work with, and require technical skills, subject matter expertise, and meaningful work to get large gains out of them.

Source and contributor context

Bridgewater Associates · Publisher

The publisher is identified here. No individual author has been inferred.

Source media

Media retrieval status: unchecked.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

Pre-training

A model's initial training on a large dataset, before further adaptation for particular uses.

NVIDIA

NVIDIA develops processors, networking and software used for AI, graphics and scientific computing.

Editorial provenance

Version 1. Publication ID: af7e0954-65bb-6383-b0cb-61be10cafaf6.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-18T12:14:10.050Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: bridge_admitted_by_model.

Observation: e69b171f22884d6b533b43903889115bdf3e6c33d5b427449929df259a5e251b. Source text hash: e67c43bd15ec28474f9ed64d35ee00d90f505014ee2e7e3893e2bb8fc2898d77.

Revision and source-check history

Version 1 · 2026-11-18T12:18:17.956Z

Greg Jensen and Jas Sekhon argue in Bridgewater Associates that Gemini 3 intensifies AI’s resource grab

On November 26, 2025, Greg Jensen and Jas Sekhon argue that Google’s release of Gemini 3 signals an intensifying AI “resource grab.” They say the model’s performance suggests scaling can continue, after challenges in training pipelines raised doubts about that approach. In their view, Gemini 3 puts pressure on other labs to invest in compute, chips and power, with major implications for capital markets and the economy. The authors also describe Google’s AI chips and hardware position as a competitive challenge to , and say that pressure helps explain Nvidia’s investments elsewhere in the AI ecosystem. They forecast that wider adoption could follow when businesses outside AI face pressure from AI-using competitors. But current models still require technical skills, subject matter expertise and meaningful work to produce large productivity gains; the authors say Gemini 3 makes that challenge more surmountable. They conclude that markets underappreciate the capital needs and economic boost over the next two years. Whether capability gains translate into productivity improvements and wider adoption remains unknown.

Original source

Bridgewater Associates · Article

Source details

Original source date: Nov 26, 2025 (2025-11-26).

Date basis: Date stated on the publisher’s page.

Added to Escape Meridian: 2026-11-18T12:18:17.956Z. First observed: 2026-11-18T01:44:02.424Z. Retrieved: 2026-11-18T01:47:58.459Z.

Limits

Text as retrieved on November 18, 2026; originally published November 26, 2025. The authors’ assessment of Gemini 3’s performance draws on external and internal testing; the cited passage does not provide test results or methods. The authors describe their estimate of Gemini 3’s compute as a rough approximation.

Open question

What evidence would show that Gemini 3’s capability gains lead to productivity improvements and wider adoption beyond the AI industry?

Source-stated

Greg Jensen and Jas Sekhon say Gemini 3’s performance indicates that pre-training scaling can continue after concerns that training challenges might slow it.

But Google cracked the challenge with Gemini 3. It is the first model released with significantly larger, effectively used pre-training compute than GPT-4o, and it showed sizable capability improvements.

Source-stated

The authors argue that Gemini 3 will add pressure on the AI ecosystem to invest in compute, chips and power, with implications for capital markets and the economy.

The path to getting better models requires more compute, more chips, and more power.

Source-stated

Jensen and Sekhon argue Google’s AI chips and hardware position challenge Nvidia and describe Nvidia’s investments elsewhere in the ecosystem as a response to competitive pressure.

This competitive pressure from Google has been a key motivating factor behind Nvidia’s extensive downstream investments in the AI ecosystem—most notably OpenAI

Source-stated

The authors say wider AI adoption could follow when businesses outside the AI ecosystem face pressure from an AI-using competitor, while noting that current models require substantial work and expertise to use for large productivity gains.

Current frontier AI models are still not easy to work with, and require technical skills, subject matter expertise, and meaningful work to get large gains out of them.

Source and contributor context

Bridgewater Associates · Publisher

The publisher is identified here. No individual author has been inferred.

Source media

Media retrieval status: unchecked.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

Pre-training

A model's initial training on a large dataset, before further adaptation for particular uses.

NVIDIA

NVIDIA develops processors, networking and software used for AI, graphics and scientific computing.

Editorial provenance

Version 1. Publication ID: af7e0954-65bb-6383-b0cb-61be10cafaf6.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-18T12:14:10.050Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: bridge_admitted_by_model.

Observation: e69b171f22884d6b533b43903889115bdf3e6c33d5b427449929df259a5e251b. Source text hash: e67c43bd15ec28474f9ed64d35ee00d90f505014ee2e7e3893e2bb8fc2898d77.

uncertain · 2026-11-18T12:14:10.050Z · anthropic/claude-opus-5.5

Publication: af7e0954-65bb-6383-b0cb-61be10cafaf6

attribution — pass: The reading correctly attributes the piece to Greg Jensen and Jas Sekhon of Bridgewater Associates. It presents the arguments as theirs throughout and does not state them as established fact.

Quick thoughts from co-CIO Greg Jensen and AIA Labs Chief Scientist Jas Sekhon on the release of Google’s Gemini 3 model

summary — pass: The reading leads with the actors and their thesis. Pre-training is defined beside the reading, and the resource grab is explained through investment in compute, chips and power. The piece's argument chain is preserved: scaling continues, investment pressure follows, Nvidia faces competition, and the Barnes and Noble adoption scenario is described. A minor fidelity gap is that the authors' forceful claim that Google is now the clear leader, and their biggest-capex-boom forecast for 2026 and 2027, are softened or omitted. The meaning stays understandable.

Google is now the clear leader in the AI race.

date — uncertain: The stated date, November 26, 2025, matches originalPublished and the dateline. However, the page carries an editor's note that it was updated on December 2, 2025, and modifiedAt is 2025-12-03. It is unclear which passages are original. Some text, such as the DeepSeek V3.2 mention, likely comes from the update. The reading presents its content as the November 26 piece without noting the revision. The reading does not rely on the clearly later passages, and it reports forecasts as the authors' views.

Editor's Note: This article was updated on December 2, 2025 to include new information.

support — pass: Each claim is source-stated and its excerpt matches: pre-training scaling continuing, pressure to invest in compute, chips and power, Nvidia's downstream investments as a response to Google, and the adoption friction of current models. Forecasts stay attributed. The closing unknown about productivity is framed as an open question, not as a claim. The limitations accurately note that the compute estimate is a rough approximation.

Our rough approximation is it appears to have used at least 2-3 times more compute than GPT-4o and GPT-5 in its pre-training

media — pass: No media was selected, and no verified candidates were supplied, so nothing is misidentified.

Challenge or correct this reading

Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.