It's an AI web, and we're just rats in the walls

The article cites Cloudflare Radar and Imperva data showing bots drive ~57–58% of HTTP requests (HTML) and ~53% of measured web traffic, respectively, alongside Pangram reporting ~1 in 4 long-form items are fully AI-generated and that LinkedIn and X are most AI-saturated. It also highlights legal allegations against OpenAI (NYT and others) that suggest training-data and output access and alleged deletion of logs in violation of preservation orders. Overall, the piece is a risk warning that AI-generated web content and summaries are degrading trust and reliability, with indirect implications for platforms and AI providers.

Analysis

The market implication is not "AI is bad for the web" so much as value migrates from content creation to traffic control and provenance. That is structurally better for NET than for ad-driven platforms: when a rising share of requests is non-human, the economic moat shifts toward bot detection, identity, and edge filtering, which can expand pricing power if enterprises view this as a security line item rather than a performance tweak. For GOOGL, the risk is second-order: if users trust AI summaries less, search engagement quality deteriorates and publishers reduce free content supply, which can eventually pressure query monetization and increase antitrust/litigation friction around training and scraping.

The near-term catalyst path is legal, not technological. Any court finding that training or output logs were mishandled would extend discovery risk, increase settlement expectations, and raise the probability of paid licensing regimes over the next 3-12 months; that is modestly positive for NYT and other premium publishers as a bargaining lever, but negative for OpenAI-adjacent infrastructure economics. Longer term, the biggest losers are low-differentiation content farms and engagement platforms where AI-assisted posting dilutes user quality; RDDT is more insulated than most because its content is already highly niche and searchable, but it still faces moderation and authenticity costs if bots keep scaling.

Consensus is probably over-moved on "AI traffic = web decay" and under-moved on the toll collectors. The more durable P&L beneficiaries are companies that can sell assurance, not creation: bot management, DLP, identity, and verified-source tooling. The thesis breaks if AI-native search materially improves citation quality and reduces bot volume without requiring incremental enterprise spend; watch for a sustained slowdown in security budget growth or any evidence that bot mitigation becomes table-stakes bundled software rather than a premium product.

Key Decisions for Investors

• Relative value: long NET / short GOOGL over 1-3 months on a basket of AI-overview and botification headlines; thesis is that traffic quality deterioration pressures search monetization more slowly than it increases demand for edge security. Stop if GOOGL proves stable CTR/ad load trends in the next ad-quarter.
• Initiate a small long NYT vs short a broad content-platform basket only if litigation discovery expands or court filings surface preserved-data issues; this is a 3-6 month optionality trade, not a near-term earnings trade.