Agentic AI is framed as the next monetization wave: business apps are moving from under 5% with built-in agents last year to 40% by year-end and 90%+ by 2030, while consumer adoption is being led by Meta’s Muse and enterprise deployment by NOW and CRM. The core claim is that agents turn AI spending into recurring revenue and compute demand, with AMZN, MSFT, GOOGL, META, NVDA, and AMD all exposed through software, cloud, and chips. The video ends with a market map: META for personal agents, NOW and CRM for enterprise workflow, NVDA and AMD for agent compute, and smaller names like INOD and FIVN for specific infrastructure slices.
Fewer than 5% of business apps had AI agents built in last year; Gartner expects that to reach 40% by end-2026 and 90%+ by 2030.
META’s Muse passed GPT as the #1 free iPhone app in less than 2 weeks and topped 3.4 million Google Play downloads, with 730,000 downloads in its first 5 days after the Sept. 8 launch.
The White House state dinner on Thursday night brought together Jensen Huang, Lisa Su, Michael Dell, Mark Zuckerberg, Satya Nadella, Sundar Pichai, Sergey Brin, Jeff Bezos, Sam Altman, Greg Brockman, and Elon Musk amid AI capex talk.
AMZN, MSFT, GOOGL, and META plan to spend about $725 billion on AI this year, up 77% from last year.
AI agents are defined as taking a goal, logging into apps, making moves, and finishing tasks on their own; use cases cited include rebooking canceled flights, customer service, help desk resets, HR requests, sales outreach, invoice matching, and coding.
Gartner says agents could be involved in over $15 trillion of B2B purchases by 2028, while KPMG found active AI-agent rollouts at big US companies rose from 11% a year ago to 54% now.
Salesforce is cited saying 70% of customer service teams using agents see results within 60 days; Gartner also expects companies to spend about $22 billion on AI features in software this year.
In one September week, Anthropic, OpenAI, META, and GOOGL released new AI models, and the claim is that each one makes agents more reliable.
IDC sees AI spending reaching $1.3 trillion by 2029, more than a quarter of all IT spend, with over a billion agents at work; McKinsey estimates agents could steer $1 trillion of US retail sales and $5 trillion worldwide by 2030.
Fortune Business Insight pegs agent-built software at about $9 billion this year, rising to $139 billion by 2034, roughly 40% annual growth.
META’s Muse launched on Sept. 8, then helped META jump 11% in one day, its best day since April 2025, and more than 25% since Muse was announced.
Zuckerberg made Muse the centerpiece at Meta Connect, calling it the path to personal superintelligence; Muse is moving into Meta AI glasses as Muse Charm, supports WhatsApp use, has outside connectors, and already includes Instacart and Shopify checkout via Shopay.
Muse is described as giving each user a secure computer in META’s cloud with its own browser, keeping work running in the background, and doing more mobile downloads in its first 12 days than ChatGPT did in its first 12 days, with 642,000 daily US users versus 231,000 for ChatGPT at the same point.
META’s pricing is free for basic use, then $20 or $100 per month for heavy users; Zuckerberg said the business model may eventually take a very small cut of transactions, potentially paid by businesses.
A 1% conversion of META’s 3.6 billion daily app users at $20 per month would be over $8.6 billion a year in new revenue, and Truist estimates Muse could generate about $28.5 billion annually by 2030 from subscriptions, shopping, and developer payments.
META’s prior AI spend hikes without visible monetization triggered selloffs: about 10% down over two days after April earnings and about 8% the next day after July earnings; Muse is presented as the first product that shows spending turning into revenue.
META is near a $2 trillion market value, after breaking out from an August range of $537-$612 above its 200-day average; the stock ran almost 40% off the August lows into the mid-$700s, with 50-day over 200-day described as a golden cross setup, $700 as first support, ~$630 as the 200-day, and ~$540 as worst-case downside back to the August lows.
OpenAI and Anthropic are described as having agents out first: Anthropic’s Claude work launched in April and its computer-use demo was in 2024; OpenAI’s ChatGPT work launched in July, both aimed more at office workflows than consumer tasks.
NOW is described as running internal IT tickets, HR requests, and approvals, with subscription revenue of about $3.9 billion last quarter, up 24.5%; AI contracts crossed $1 billion a year, new AI deals grew more than 40% sequentially, agents running on NOW rose ninefold in 9 months, and its AI control tower passed 500 customers in six months.
CRM’s Agent Force is generating over $1.5 billion in recurring revenue, up more than 240%, while agents completed 3.2 billion tasks this quarter, almost double the prior quarter; CRM raised full-year revenue outlook to $46 billion and launched Headless 360 so agents can use CRM without screens and buttons.
NVDA posted $96.2 billion in revenue last quarter, more than double a year ago; data center revenue was $89 billion, up 117%, and next-quarter guidance is about $108 billion as agents drive more compute demand.
NVDA says agent compute can require 15 to 100 times the compute of a human using the tool; it unveiled Vera, called the first CPU built for AI agents, with CPU sales already above $5 billion over the last year, and Vera Rubin is now in full production with management expecting about 70% growth next fiscal year.
AMD posted record revenue of $11.5 billion, up 50%, with data center up to $6.7 billion and server CPU sales up roughly 70%; Lisa Su says agents could shift data centers from one CPU for every 4 to 8 AI chips toward one-for-one, and AMD has deals with OpenAI and META plus a 2-gigawatt Anthropic order for its newest AI chip starting in 1H27.
AMD says data center sales can more than double in 2027, AI chip business can grow well over 100%, and the AI chip market could reach about $1.4 trillion by 2030.
INOD posted record revenue of $92.1 million last quarter, up 58%, kept about 49 cents of every dollar after direct costs, doubled profit to $14.4 million, and ended with about $250 million in cash and no debt.
FIVN’s AI revenue grew 78% to about 15% of subscription revenue from 9% a year ago; it closed about a $100 million deal with a Fortune 100 financial firm via GOOGL’s cloud marketplace, launched Voice AI agents, and raised AI growth guidance to at least 60% for the year.