
Investors should consider Cloudflare (NET) as a primary play on the "Agentic Web," as its Isolate technology and VoidZero acquisition position it to handle the massive CPU demands of AI agents more efficiently than traditional cloud providers. The company’s partnership with NVIDIA (NVDA) to deploy GPUs in over 350 cities makes it a leader in "Edge Inference," allowing AI processing to happen locally and bypass data center latency. NVIDIA remains a high-conviction hold as demand shifts from centralized training clusters to this distributed edge infrastructure. To capitalize on the "Token Economy," look for companies that reduce AI processing costs through data efficiency, such as Cloudflare’s automated conversion of web content into AI-friendly Markdown. Finally, prioritize infrastructure investments in "distributed" models that utilize existing buildings to bypass the power grid constraints and long lead times facing massive new data center projects.
• Acquisition of VoidZero (Vite): Cloudflare recently acquired VoidZero, the company behind Vite, a massive developer platform with 130 million weekly downloads. The goal is to integrate Vite closely with Cloudflare’s developer platform to make it the "default" and best place to run Vite-powered projects. • AI Agent Infrastructure: CEO Matthew Prince predicts that AI agents (autonomous programs performing tasks) will soon outnumber humans 1,000-to-1 on the internet. • Architectural Advantage: Traditional cloud containers (like those from AWS or Google Cloud) are too "heavyweight" for the massive CPU/GPU demands of millions of AI agents. Cloudflare uses "Isolates," a much more efficient technology that allows more agents to run on the same hardware. • Traffic Milestone: Bot/agent traffic surpassed human traffic on Cloudflare’s network in the first half of 2026 (noted as happening much sooner than the original 2027 prediction). • Inference at the Edge: Cloudflare has partnered with NVIDIA to put GPUs in over 350 cities worldwide. This allows AI "inference" (the processing of AI requests) to happen within milliseconds of the user, rather than traveling to a distant data center. • Cost and Efficiency: Cloudflare claims to be more cost-effective than competitors because they are a network first; bandwidth is effectively free for them, and they often place equipment in existing partner facilities, saving on power and real estate costs.
• Bet on the "Agentic" Web: Cloudflare is positioning itself as the essential plumbing for the next phase of the internet, where AI agents do the browsing, researching, and buying instead of humans. • Efficiency as a Moat: As AI scaling hits power and hardware bottlenecks, Cloudflare’s Isolate technology and edge computing model offer a more sustainable and cheaper alternative to traditional hyperscalers like AWS. • Sovereign AI Trend: Investors should note the "sovereign" desire in regions like Europe to keep data local. Cloudflare’s distributed footprint (350+ cities) makes it a primary beneficiary of regulations requiring data to stay within specific borders.
• Edge Partnership: Cloudflare confirmed a deep partnership with NVIDIA to deploy GPUs across their global edge network. • Hardware Demand: The transcript highlights that while "frontier" models run in massive data centers, a significant portion of AI tasks will move to the edge for latency and privacy reasons, sustaining long-term demand for NVIDIA hardware outside of just the major "hyperscaler" data centers.
• Diversified Demand: NVIDIA’s growth is not just tied to massive AI training clusters but also to the "inference" layer at the edge of the network (where Cloudflare operates).
• Shift in Web Consumption: The web is moving from "human-readable" (HTML) to "agent-readable" (Markdown/JSON). Cloudflare is already automatically converting content to Markdown to save "tokens" (the currency of AI processing) and make it easier for agents to digest information. • Exponential Web Growth: After a plateau since 2015, the number of websites is growing exponentially again as AI agents create and consume content. • Bottleneck Risks: The sheer volume of AI agents could require 30-40 times the current global capacity of CPUs and GPUs if built on old architecture.
• Efficiency Tools are Key: Companies that help "squeeze every ounce of performance" (like those specializing in Markdown conversion or efficient coding platforms) will be high-value targets. • Token Economy: Investors should watch for companies that reduce "token costs." Lowering the amount of data an AI has to process is as valuable as increasing the speed of the processor itself.
• Decentralization vs. Centralization: While Google, Microsoft, and AWS build massive "hyperscale" facilities, Cloudflare represents a "distributed" model—smaller clusters of machines in thousands of existing buildings. • Resource Concerns: The CEO dismissed some environmental concerns (like water usage) as "silliness" but acknowledged that power grid capacity is a real constraint.
• Infrastructure Optionality: Investors should look for companies that aren't just building new "megawatt" facilities but are finding ways to utilize existing "brownfield" data centers and ISP buildings to avoid the long lead times of new construction.
• Public vs. Private: Matthew Prince argues that being a public company is "healthier" than being private because it allows for "no-fault divorce"—investors can sell if they disagree with management, creating a more honest relationship than the "dysfunctional" lock-in of Venture Capital. • The "IPO Pop": Cloudflare intentionally priced its IPO to see a 20% "pop" on the first day to build long-term relationships with institutional shareholders, rather than trying to extract every penny of value at the start.
• Management Quality: When evaluating upcoming IPOs, look for founders who prioritize "long-term shareholders" over "maximizing the opening price." A stable top-10 shareholder list is a sign of a healthy public company.

By John Coogan & Jordi Hays
Technology's daily show (formerly the Technology Brothers Podcast). Streaming live on X and YouTube from 11 - 2 PM PST Monday - Friday. Available on X, Apple, Spotify, and YouTube.