Hunting for Trade Ideas with Google's AI Roadmap

Google’s Project Genie dropped last week and sent shockwaves through the gaming sector.

Unity crashed over 30% in days. The market suddenly realized that AI can now generate playable, interactive worlds from a single prompt.

Here’s the thing though…

Google literally told you this was coming.

They published the Genie research paper in February 2024. It described an “unsupervised generative interactive environment” trained on internet video. The capability was disclosed nearly two years before the product shipped.

Investors weren’t paying attention.

This is a recurring pattern with Big Tech AI (especially Google's philosophy).

Research gets published. Previews get announced. Capabilities get telegraphed months (sometimes years) in advance. And then when the product drops, everyone acts surprised.

So I asked myself: what else is on Google’s public roadmap that investors are missing?

Google publishes a ton of signal if you know where to look:

• ArXiv papers from DeepMind
• Model changelogs and deprecation notices
• “Preview” to “GA” transitions in Vertex AI
• Subscription entitlement pages and billing start dates

The problem is no human can reasonably track all of this and map it to investable conclusions.

This is where AI comes in.

I’m going to show you how I used Deep Research to systematically mine Google’s public AI roadmap and translate it into a watchlist of tickers that could be affected by upcoming capability rollouts.

Here’s the plan:

1. Prompt 1 uses Deep Research to map Google’s disclosed capabilities, forecast rollout timelines, and identify leading indicators
2. Prompt 2 takes those capability forecasts and maps them to specific tickers (long and short opportunities)

The Prompt

We're using ChatGPT's Deep Research function here for the first prompt then GPT thinking mode for the second prompt. The first prompt gets us the roadmap prediction via raw sources and the second prompt gives us an excel sheet with tickers that would be impacted by this.

Prompt 1: Find Gemini New Capability Roadmap

​Google_Roadmap_Prompt.pdf​

​

Prompt 2: Map Roadmap to Tickers (GPT Thinking Mode)

​Google_Ticker_Mapping_Dave_Wang.pdf​

The Result

Deep Research Roadmap Output: Forecasting Google AI Product Rollouts as of February 4, 2026.pdf​

Ticker Mapping Output: Link​Google_AI_Roadmap_Ticker_Impact_Map.xlsx​

​

Our AI analysis systematically mapped Google's public research artifacts, API changelogs, and product documentation into a probabilistic rollout forecast with direct ticker implications.

Here are the most likely features in Google's AI roadmap based on research and public data...

The key insight? Google is converging on "agentic distribution." They're embedding agents into every consumer surface (Search, Chrome, Gemini app) and gating access via subscription tiers. This is the playbook.

Ticker Impact Summary (GPT's Flags):

A few things jumped out:

• NVDA appears in the top 5 of almost every capability cluster. Not surprising, but the AI confirmed that world-model inference (Project Genie) and agentic workloads are compute-intensive enough to keep accelerator demand strong even as cost curves compress.
• RDDT and TRIP scored highest reaction risk on the short side. Chrome auto-browse and Search AI Mode directly threaten their traffic economics. If agents summarize content and execute bookings without clicking through, these businesses lose their moat.
• PATH (UiPath) and ASAN (Asana) show up repeatedly in the "workflow automation negative" bucket. The logic: if Google ships cloud-native agent primitives at scale, standalone RPA and task management tools face substitution risk.
• Unity and Roblox are genuinely mixed. The bull case is more creators feeding engine usage and UGC platforms. The bear case is world-model experiences that bypass traditional toolchains entirely. AI flagged both directions with high reaction risk

Obviously don't blindly trust these results - you need to do proper due diligence on each name. This is just for idea generation purposes.

The Structural Takeaway:

Google's research-to-product pipeline follows a consistent pattern:

1. ArXiv paper (capability disclosed)
2. Limited research preview (small cohort access)
3. Labs prototype (tier-gated, constrained)
4. Developer preview (API model IDs appear)
5. Enterprise GA (billing starts)
6. Broad consumer rollout

If you were to zoom out, you can utilize the same technique we did here to any company whose product roadmap creates massive market moves like Google's did.

All you need to do is reverse engineer where the signal is coming from then run the right prompts!

Personal

I've been on the road for the last 3 weeks - I'm curious where my most active readers are located in the world.

Hit the reply button and let me know!

I'm expecting to travel a lot between US East Coast, Asia, and Middle East this year.