Cloudflare CEO Says the Internet Is Changing Exponentially

Cloudflare said AI-agent web traffic surpassed human activity on its network in May and expects agentic traffic to grow to ~1,000x human levels within five years, though analysts on the show disputed the math. The episode also highlights U.S. government mining-related funding (reported as $3B, including $100M education and a $1.4B DOE loan for a lithium-ion battery startup) with skepticism on near-term mining-stock profitability. Intel announced a $15B equity raise to fund capex and “physical AI,” characterized as a dilution-light move (stated as <3% dilution at current market value) versus potentially higher future costs of capital.

Analysis

Agentic traffic matters less as a headline and more as a shift in who captures value on the web. If machine-to-machine requests keep rising, the open internet becomes a lower-quality ad medium and a higher-value infrastructure problem: verification, bot management, edge routing, and cost-efficient inference all get monetized before content does. That is structurally favorable for NET; it is structurally hostile to impression-based ad models like TTD, where the risk is not a demand collapse but a slow degradation of the addressable human inventory and pricing power over the next 6-18 months.

The mining-policy angle is a classic timing mismatch. Even if the policy intent is correct, education spending is a 4-5 year pipeline fix while the actual bottlenecks are permitting, social license, and commodity-cycle discipline. The market should fade any reflexive rally in pre-revenue or story-driven mining equities: if new capital chases “strategic minerals” without visible cash flows, the likely outcome is dilution rather than production. The second-order beneficiaries are more likely to be industrials with actual equipment/service exposure than miners themselves, but that payoff is too indirect for a clean trade today.

Intel is the cleaner setup because the raise changes the left tail before it changes the earnings model. Dilution at a rich multiple is manageable; a future liquidity event at a lower multiple would be far worse, so the financing itself de-risks the equity even if it doesn’t yet prove the growth story. The key falsifier is simple: if foundry/external-wafer momentum and gross margin trajectory do not improve over the next 2 quarters, the market will treat this as bridge financing for a still-unproven capex cycle, and the stock should give back the post-raise optimism.

Key Decisions for Investors

• Long NET / short TTD for 1-3 months: market the spread as a bet on agentic traffic monetization vs. ad-inventory degradation. Risk/reward improves on any selloff in TTD or if NET confirms bot-mitigation demand in the next earnings print.
• Buy INTC on post-offering weakness for a 6-12 month horizon: the raise lowers distress risk and extends runway for foundry optionality. Falsify if next two quarters fail to show improving foundry bookings or margin inflection.
• No chase in speculative mining names; treat the policy news as a watch item, not a trade. Reassess only if permitting reform or project-level cash flow turns visible within 6-18 months.
• If long AI infrastructure already, trim the highest-multiple names and rotate toward self-funding infrastructure cash generators. The funding environment can stay hot, but a single financing scare would compress multiple-sensitive winners first.