Date: August 17, 2026

The most important thing to know about today’s AI market is not which model is smartest this week. It is which companies can secure enough compute to meet demand at scale without blowing up unit economics.

That is why this Axios headline matters: citing Bloomberg figures, Axios reported today that Anthropic’s annualized revenue run rate has moved above $65 billion ahead of an expected IPO. Whether you take the number literally or treat it as directional, the message is the same. AI demand is behaving like an industrial-scale business, not a lab experiment.

But investors have a practical constraint. You cannot buy Anthropic yet. So the question becomes the one Trading Stocks Now is built around: if you could buy only one stock today to express this theme, which one offers the best blend of business quality, momentum, catalysts, and risk versus reward?

Our answer is straightforward.

One stock to own right now: Broadcom (AVGO).

Why This Stock Now

When AI markets are euphoric, investors chase application-layer names. When AI markets get disciplined, they focus on what is scarce. Today, scarcity is still compute capacity and the supply chain that delivers it.

Anthropic itself has been explicit about moving toward public markets. On June 1, 2026, the company said it confidentially submitted a draft registration statement on Form S-1 to the SEC. If the IPO window opens, it will not just be a “hot deal.” It will be a forcing function for the market to anchor AI valuations to real revenue and real cost structure.

The cleanest public way to express that debate is to own the capacity trade, not the lottery ticket.

The Business

Broadcom is a diversified semiconductor and infrastructure software company. For this issue, the focus is its role in scaling specialized compute supply chains, including Google TPU-related capacity delivered through Broadcom, as well as the adjacent connectivity components that matter when AI clusters get bigger and more expensive.

The way to think about Broadcom in this cycle is not as a “chip story.” It is as an industrial partner. AI leaders want predictable delivery, tighter integration, and a roadmap that extends multiple generations. If you believe frontier models are becoming the interface for work, the infrastructure has to expand for years, not quarters.

Why Wall Street Is Paying Attention

Institutional investors look for proof points. The best proof is disclosed commitments that attach real scale to real counterparties.

On April 6, 2026, Broadcom filed an 8-K describing an expanded collaboration involving Broadcom, Google, and Anthropic. The filing said Anthropic, beginning in 2027, will access through Broadcom approximately 3.5 gigawatts as part of multiple gigawatts of next-generation TPU-based AI compute capacity committed by Anthropic.

This matters for one reason. It reframes the investor conversation from “AI is popular” to “AI is being provisioned like power.” Once you start talking in gigawatts and multi-year access, you are no longer debating whether demand exists. You are debating who captures the economics of delivering it.

What’s Driving the Opportunity

• A revenue curve that is forcing capex decisions: Axios reported today that Anthropic’s annualized revenue run rate is above $65 billion, citing Bloomberg figures.
• Named-customer, disclosed scale: Broadcom’s April 6, 2026 8-K describes Anthropic accessing approximately 3.5 gigawatts of next-generation TPU-based compute beginning in 2027.
• A second-order tailwind: As AI vendors move from experimentation into enterprise workflows, reliability and cost per unit become the differentiator. That dynamic tends to support scaled supply chains and long-duration infrastructure budgets.

Put differently, the bullish case is not “AI is the future.” The bullish case is that AI has already become a procurement cycle.

What Could Go Wrong

Investors should take the bear case seriously because the trade is not free.

• Valuation risk: AI beneficiaries are widely owned, and sentiment can swing faster than fundamentals.
• Capex digestion cycles: Even in strong secular trends, there are periods where customers pause to absorb what they have built.
• Customer concentration and architecture shifts: If large buyers shift workloads, switch silicon strategies, or compress spend, the group can re-rate quickly.
• Timing risk: The 2027 start for the next-generation TPU capacity supports duration, but it can create near-term impatience if investors demand immediate revenue translation.

Thesis invalidation: credible evidence that frontier labs are materially reducing multi-year capacity commitments, or that Broadcom’s role in the TPU supply chain is structurally shrinking rather than scaling.

The Bottom Line

Anthropic’s reported $65 billion run-rate headline is not a reason to chase every AI name. It is a reason to tighten the frame. If AI revenue is scaling that fast, compute capacity stays scarce, and the economics accrue to the companies that can deliver scaled, multi-year infrastructure.

That is why Broadcom (AVGO) is the single most compelling stock to focus on today. It is a public, liquid, institution-friendly way to express the next leg of the AI buildout, anchored by a disclosed, gigawatt-scale relationship that directly involves Anthropic and Google’s TPU roadmap.

Action: Treat AVGO as the capacity trade behind the Anthropic IPO cycle. If the market keeps rewarding measurable AI demand, this is where the debate is likely to concentrate.