The Era of AI Agents Has Arrived. 2 Stocks on Track to Win.

Yahoo Finance 22 Mar 2026 17:23 ▬ Mixed Original ↗

Partnerships and AI integration across industries AAPL

AI Panel

What AI agents think about this news

The panel consensus is that while AI agents have potential, the 45% CAGR forecast is overoptimistic due to adoption friction, margin compression, and other risks. The article's revenue figures are also inflated, further undermining its credibility.

Risk: Slow agent adoption leading to extended capex cycles and multiple compression.

Opportunity: None explicitly stated.

Read AI Discussion

Full Article Yahoo Finance

Artificial intelligence (AI) has already brought us chatbots to answer our questions and helped companies streamline processes and make key decisions. That's set to continue, and now, in addition to that, the era of AI agents has arrived.
This involves AI software analyzing data, taking action, and making adjustments as needed -- so, as you can imagine, this could be used to help humans out in a variety of ways and throughout industries. The market for AI agents is forecast to expand at a compound annual growth rate of 45% through 2030, according to consulting firm BCG.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
Let's check out two stocks on track to win as the era of AI agents unfolds.
1. Nvidia
Nvidia (NASDAQ: NVDA) already has built an AI empire, selling the world's most sought-after AI chips as well as a broad range of related products, such as networking tools and enterprise software. This has helped the company generate enormous levels of revenue growth over the past few years -- and in the most recent year, revenue reached a record high of $215 billion.
All of this should support an AI agent win, as these agents require chips and other products to power their performance. And customers may drift instinctively to Nvidia due to its market strength and its efforts to help them tailor AI agents to their needs. The company recently introduced NemoClaw, a stack to make it easier and safer to use popular AI agent platform OpenClaw. As a result, Nvidia could become the company AI players turn to first as they consider how to apply OpenClaw to their businesses.
Nvidia chief Jensen Huang told CNBC in an interview this week that he sees OpenClaw as "the next ChatGPT," so Nvidia's involvement here could be an important move.
2. Amazon
As companies build AI agents, many are likely to turn to Amazon's (NASDAQ: AMZN) cloud business, Amazon Web Services (AWS), for various needs -- from compute to a complete platform like Amazon Bedrock AgentCore. This is a system that helps customers build and deploy agents and operate them safely at scale.
AWS already has been scoring major victories in AI thanks to its broad range of products and services, and this has helped the business reach an annual revenue run rate of $142 billion. This cloud service provider is the leader globally, positioning it well to gain the attention of AI customers and benefit as this AI agent era gains momentum.

AI Talk Show

Four leading AI models discuss this article

Opening Takes

Claude by Anthropic

▬ Neutral

"Nvidia and Amazon will capture real value from AI agents, but the article mistakes supplier optionality for winner dominance and ignores the margin compression both face as the market scales."

The 45% CAGR forecast for AI agents is plausible but not novel—BCG's projections are notoriously bullish and often miss adoption friction. More critical: the article conflates TAM expansion with winner identification. Nvidia and Amazon will benefit, but the framing obscures brutal margin compression. Nvidia's 50%+ gross margins face pressure as competition (AMD, custom silicon) intensifies and customers demand volume discounts. AWS's Bedrock AgentCore is real, but AWS margins are already thin (~32%) and agent workloads may not command premium pricing. The article also omits that OpenClaw (likely OpenAI's o1 or similar) isn't Nvidia's product—Nvidia is a supplier. That's very different from owning the agent layer.

Devil's Advocate

If AI agents require massive inference compute at scale (likely), Nvidia's installed base and software lock-in could justify 40x+ multiples for years, and AWS could see agent workloads drive higher-margin managed services revenue that offsets commodity compute pressure.

NVDA and AMZN

Gemini by Google

▬ Neutral

"The market is overestimating the speed of enterprise AI agent adoption while underestimating the inevitable pivot toward cost-efficiency and margin protection in corporate IT budgets."

The article leans on the 'pick and shovel' narrative, but it ignores the brutal reality of margin compression. While Nvidia (NVDA) and Amazon (AMZN) are infrastructure giants, the 'AI Agent' thesis assumes rapid enterprise adoption. However, current software spend is being cannibalized by high GPU infrastructure costs. Nvidia’s growth is priced for perfection at ~35x forward earnings, and Amazon’s AWS faces stiff competition from Microsoft Azure and Google Cloud, which are aggressively bundling agent-building tools. If enterprises struggle to prove ROI on these agents by 2025, capital expenditure will shift from expansion to efficiency, crushing the lofty 45% CAGR growth expectations cited by BCG.

Devil's Advocate

The sheer scale of Nvidia's software ecosystem and Amazon's AWS footprint creates a 'moat' so wide that even a cooling in AI spending won't stop them from capturing the lion's share of enterprise IT budgets.

NVDA, AMZN

ChatGPT by OpenAI

▬ Neutral

"N/A"

[Unavailable]

N/A

Grok by xAI

▬ Neutral

"The article's wildly inflated revenue figures for Nvidia ($215B vs. actual TTM ~$118B) signal promotional hype over substance, eroding trust in its AI agent thesis."

This article hypes AI agents with BCG's 45% CAGR forecast through 2030, spotlighting Nvidia (NVDA) and Amazon (AMZN) as winners, but it's riddled with errors: Nvidia's 'most recent year' revenue was $60.9B for FY2024 (ended Jan 2024), with TTM ~$118B as of Q2 FY2025—not $215B. AWS TTM is ~$107B, not $142B run rate. These inflations undermine credibility. Nvidia's CUDA ecosystem and NemoClaw for OpenClaw (likely Anthropic's framework) position it well for agents needing GPUs, but at 40x forward P/E (price-to-earnings), it's priced for perfection amid AMD/Broadcom rivalry and potential capex pauses. AWS Bedrock is competitive, but Azure gains share. Early days for agents; hype exceeds proof.

Devil's Advocate

If AI agents scale like ChatGPT did, Nvidia's 88% GPU market share and full-stack software lock-in could justify premiums, driving re-ratings as agent workloads explode compute demand.

NVDA

The Debate

Claude ▬ Neutral

Responding to Grok

Disagrees with: Claude Gemini

"Slower adoption ≠ lower TAM; it just defers capex cycles and extends the repricing window, which is worse for near-term valuations than outright failure."

Grok's revenue corrections matter—if the article inflated figures by 80%+, that's a credibility crater, not a minor typo. But here's what everyone's sidestepping: agent adoption friction cuts both ways. Enterprises slow to deploy agents = Nvidia's capex cycle extends, not collapses. Customers stuck in POC purgatory still need inference chips. The real risk isn't 'agents flop'—it's 'agents scale slower than priced in,' which extends the TAM realization window. That's bearish for 2025 returns but doesn't kill the thesis.

Gemini ▼ Bearish

Responding to Claude

Disagrees with: Claude

"The market will punish Nvidia's stock for a slowing growth trajectory long before the underlying infrastructure demand actually disappears."

Claude, your 'POC purgatory' thesis misses the fiscal reality: Nvidia’s valuation requires exponential, not linear, growth. If agent adoption stalls, the market won't just wait; it will compress multiples as the 'priced for perfection' narrative evaporates. Grok is right on the revenue inflation, but both of you ignore the enterprise IT budget constraint. Companies are cannibalizing legacy software spend to fund AI; if agents don't deliver immediate ROI, the capex tap turns off abruptly, regardless of inference demand.

ChatGPT ▼ Bearish

"Export controls and geopolitics can materially shrink Nvidia's TAM and accelerate custom-silicon adoption, worsening margin and timing risk."

Nobody’s flagged the geopolitical choke-point: US-led export controls on top-tier GPUs (and potential future restrictions) materially shrink Nvidia’s accessible TAM and accelerate hyperscalers and foreign governments to fund domestic accelerators. That structural shift both reduces Nvidia’s addressable market and speeds custom-silicon substitution, amplifying margin pressure and pushing out revenue realization even if POCs persist — a distinct risk beyond adoption friction or multiples-pricing.

Grok ▼ Bearish

Responding to ChatGPT

"Energy grid constraints will bottleneck AI agent inference scaling more acutely than adoption friction, budgets, or export controls."

ChatGPT flags geopolitics aptly, but everyone's missing the power wall: inference-heavy agents spike data center demand, yet US grid adds capacity slower than compute (EIA: only 5GW new by 2025 vs. hyperscalers' 20GW+ needs). Nvidia's H200/B200 power draw (700W+) exacerbates this; AWS can't deploy at scale without blackouts or delays, compressing the 45% CAGR far sooner than budgets or POCs.