Google's been previewing Gemini 3, and the demos are impressive.

Someone one-shotted a simple version of Windows OS.

Another user composed and played a piano song.

Look at this neobrutalist website Gemini coded in one go.

From a user perspective? This looks very promising.

But what does this mean for $GOOG investors?

The real question: is this a step-change model that moves the needle on Google Cloud and Workspace adoption? Or just another incremental upgrade with flashy demos?

Here's the problem: Gemini 3's launch is playing out entirely on X.

No official technical report. No clean benchmark suite. Just scattered threads, leaked previews, and developer reactions.

Traditional sell-side research won't cover this for weeks. By then, the edge is gone.

I used Grok to evaluate Gemini 3's commercial impact in real time.

Here's why this works: Grok has native X access. We can scrape developer demos, benchmark leaks, and adoption signals before they hit official docs or analyst reports.

The plan:

Scrape official Google announcements + high-signal X threads
Compare Gemini 3 vs 2.5 on coding, reasoning, multimodal
Map product rollout (Cloud, Workspace, Android)
Assess revenue impact

Was this email forwarded? Subscribe to get more AI investing breakdowns just like this!

Subscribe Here

The Prompt

Use this prompt only in Grok (Elon's AI model on X)

This is since only Grok has access to X posts - most of the information we're looking for on Google's new model is rumor sitting on X.

Download prompt: gemini_3_research_prompt.pdf

The Result

Full output here: Link

Here's what Grok surfaced after scraping X + official sources:

Gemini 3 is a solid upgrade, not a revolution.

The demos look impressive. But when you dig into the actual performance data, Gemini 3 improved (allegedly) ~2-3% across most benchmarks vs Gemini 2.5. It's still trailing GPT-5 and Claude 4 on core reasoning tasks.

Where it actually wins: coding UI and reading documents.

X threads show developers trying out the new Google model for frontend work. Gemini 3 can one-shot full UI builds better than 2.5. Claude has generated significant revenue despite a fraction of OpenAI's userbase due to deeper penetration into enterprise.

For $GOOG investors this matters because these are revenue-generating Cloud workloads.

Three places Google monetizes this:

Cloud: Early pricing signals at $0.30/M tokens. Enterprises report 20-30% faster analytics tasks. Could pull workloads from AWS via Gemini share gain.
Workspace: No-code agents in Docs/Sheets/Gmail. If adoption hits, this drives higher-tier subscription upgrades.
Android: On-device image/video gen via Nano Banana 2. Competes with Apple Intelligence for creative features.

The catch: all the numbers are provisional.

No official benchmarks yet. Everything circulating on X is from leaks, previews, or marketing claims. The real model card drops later in the year.

Bottom line: Gemini 3 closes the gap but uncertain if the improvement is good enough for significant share gains for enterprise.

I will say however Google has incredible advantages in the AI race and my horse to win in the long run for foundation model performance....maybe that's for a different thread.

New Resource!

Check it out here

I created an email based 5-day crash course on prompting for finance.

And the best part is it's free.

Every day, you'll get an email including a short video, PDF handouts, and my written thoughts.

Here's the curriculum:

Day 1: How to prompt like a pro
Day 2: Prompt techniques for finance
Day 3: Advanced prompting techniques
Day 4: When should you use each AI tool?
Day 5: How Wall Street pros are actually using AI today

Once you enroll here reply back to this email to let me know you're in. Helps me make better content if I know what the people want!

When you're ready, here's how we can work together:

Join Wall Street Prompt - On-demand AI learning
Live Corporate Training - Live AI for finance workshops

Google Previewed Gemini 3. I Used AI to Separate Signal from Noise.