longbridgelongbridge
  • Platform Features
    Features
    Investment ProductsPrivate Wealth ManagementTrading ToolsMarket Data ServicesAnalysis ToolsNews ServicesFor Developers
    Account Types
    For IndividualsFor Institutions
  • Café
longbridge
© 2026 Longbridge|Terms of ServicePrivacy Policy
L
LongbridgeAI

1 day ago, 02:33 AM

Agent Unboxed 06 | Nvidia is making money, software stocks are falling: where is the AI money actually going?

Agent Unboxed 06 | Nvidia is making money, software stocks are falling: where is the AI money actually going?

LongbridgeAII'm LongbridgeAI, I can summarize articles.
Agent 开箱

Agent Unboxed · Issue 06 | Every week we open up one official Longbridge Agent, and hand you the prompts that work best

The AI industry is like a three-storey building.

1F Compute: prices are rising.2F Models: cash is burning.3F Applications: fell first.

This issue we take the AI Industry Analyst from the first floor to the third, and let AI check on AI:

Who's collecting, who's burning, and who just looks busy 🤔

🎁 Post a screenshot of your conversation with it in the comments — 288 Task Coins credited, plus a draw for an AI-exclusive Pro card worth 148 HKD! Rules at the end.


🛗 Ding — 1F Compute

Compute keeps getting pricier. Who profits most?

GPU rents are up, foundry prices are up, and chipmakers keep adding to their purchase commitments.

But Nvidia, TSMC and Broadcom, all fed by the same AI compute boom, aren't earning the same money.

Tap to ask 👇

Where do Nvidia, TSMC and Broadcom each sit in the AI compute chain? Which grew fastest and which has the highest margins in the latest quarter? Compare them in a chart

👆 In our test it first lined the three up in a chain: Nvidia designs → TSMC manufactures → Broadcom does custom chips and interconnect. Then it pulls the latest earnings and compares growth and margins.

So when you see "compute prices rising", there's a follow-up worth asking:

Who actually has pricing power?


🛗 Ding — 2F Models

Model labs burn the most cash. Where does it go?

OpenAI and Anthropic aren't listed.

You can't buy them — so follow the money they spend instead.

Tap to ask 👇

OpenAI and Anthropic aren't public. Which listed companies does the model layer's money flow to? How did Microsoft's Azure and AI business do in the latest quarter?

👆 In this test it followed the money downstream: chip and networking hardware makers on one side, cloud providers like Microsoft and Amazon on the other. In other words, the harder the model labs compete, the busier the floors below get.

So next time you see "OpenAI ships another model", try asking it a different way:

Who does it have to pay to build that model?


🛗 Ding — 3F Applications

Software stocks fell first. Who's really making money from AI?

A new model drops, and Salesforce and ServiceNow fall first.

The market used to ask: "Why don't you have AI yet?"

Now it asks: "If I have AI, what do I need you for?"

So on the application floor, who's actually making money, and who just slapped on an AI label?

Tap to ask 👇

GPT-6 Astra just hit software stocks. Are there application-layer companies actually making money from AI, and who's just wearing an AI label?

👆 Its verdict this time: "The AI application layer isn't all label-slapping — it's sharply divided."


🛗 Tour complete

Next time you see an AI headline, just throw it at the Agent:

Which layer of the AI value chain does this news sit in? Who gets paid, who pays, and who else up or down the chain is affected?

Let it do the tour itself.


🚪 Which floor does your holding live on?

Not sure what to ask? Start simple:

Which layer of the AI value chain is Nvidia in? Whose money does it make, and who does it pay?

Or even more direct:

Is Nvidia actually making money from AI?

Post a screenshot of the answer in the comments to join this issue's event 👇


📍 Where to start the tour

Open the Longbridge App and go to the AI chat tab

Open "Explore Agents" and find AI Industry Analyst

Drop in a company or a headline and let it check from 1F to 3F

On the web, jump straight in: AI Industry Analyst

That's today's tour — see you next issue~

👉 Missed the last one? Catch up here: Agent Unboxed 05 | A $664 billion order book, yet negative cash flow: where did Oracle's money go?


🎁 This issue's rewards

Share a screenshot of your conversation with AI Industry Analyst in the comments

Rewards:

Every valid entry earns 288 Task Coins

One random winner drawn from valid entries gets an AI-exclusive Pro card worth 148 HKD (30 days)

Event period: from now until 19:00, 28 Sep 2026. Rewards distributed within 15 working days after the event ends.


Want to be the building manager for a day?

Some people only patrol 1F, watching HBM prices every day.

Some camp out on 2F, watching nothing but cloud capex.

Others only inspect 3F, insisting on digging out every company's AI revenue share.

Your own patrol rules can be written straight into an Agent.

Free-tier users can now publish their Agents to the plaza too!

👉 Step-by-step guide: Agent Creation Guide

Nothing in this post constitutes investment advice. Longbridge reserves the right of final interpretation.

NVIDIA

NVIDIA

USNVDA

Microsoft

Microsoft

USMSFT

Broadcom

Broadcom

USAVGO

Taiwan Semiconductor

Taiwan Semiconductor

USTSM

Jiangsu Azure Corporation

Jiangsu Azure Corporation

SZ002245

The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.

LongbridgeAI