06-reference

innermost loop bots pass humans on the web

2026-06-04·reference·source: Innermost Loop·by Alex Wissner-Gross

"Welcome to June 4, 2026" — Alex Wissner-Gross

Why this is in the vault

Daily frontier-AI digest, written as woven prose rather than bullets. The lead frame — the Singularity is "creeping toward the edge," with modality after modality falling to open models — is on-theme for the agent-capability race RDCO tracks, and it lands a concrete milestone: bots have passed humans in web traffic for the first time, per Cloudflare's Matthew Prince. The silicon/capital thread (TSMC's multi-year supply-shortage warning, ~$725B in 2026 hyperscaler AI spend, SpaceX's $55B Terafab, the EU Cloud and AI Development Act) is live datapoint material for the chip-fab/memory capital-cycle investing thesis. Secondary threads on agent-driven commerce, "answer engine optimization," and the dual-use biosecurity letter are adjacent to RDCO's harness practice and its bot-economy awareness.

Issue contents

Curation digest, ~25 linked items strung into a single thematic narrative across roughly eight movements. Self-promo: standard Substack subscribe CTA only; not sponsored content. All deep links are Substack redirect wrappers pointing to third-party news sources; no affiliate or self-product links. No sponsor block. Closing line riffs on Hamlet — more in heaven and earth than is dreamt of "in your training data."

Mapping against Ray Data Co

Strong. The issue's organizing arc — modality after modality (vision, audio, video, voice) falling to open-weight, laptop-runnable models, capped by the bots-pass-humans-on-the-web milestone — is a sharp restatement of the commoditization-of-raw-capability read that underpins RDCO's harness-engineering thesis: when frontier generation runs on a laptop under Apache 2.0, the durable moats move to control, memory, and orchestration, not raw model access (per [[2026-06-01-innermost-loop-memory-worth-more-than-oil]] and [[2026-05-11-innermostloop-harness-eats-the-model]]). The "bots now outnumber humans on the web" datapoint is the demand-side mirror of that thesis — and the Meta Business Agent and "answer engine optimization" items are concrete, near-term shape of the agent-economy RDCO's COO build is itself an instance of.

Two threads are direct investing-thesis inputs. (1) The silicon/supply movement — TSMC's multi-year supply-shortage warning, the ~$725B 2026 hyperscaler AI spend figure, SpaceX's $55B Terafab, Foxconn/Intel Xeon rackscale — is on-thesis corroboration for the chip-fab/memory capital cycle the founder places RDCO in Phase 2 of; Wei's explicit pledge to avoid memory-style abrupt price hikes is a notable signal for the memory-cycle leg specifically. (2) The capital-side items (Alphabet's $80B raise, SpaceX's ~$1.77T IPO, the EU Cloud and AI Development Act tripling European DC capacity) are the financing-and-policy side of that same buildout maturing. Caveat: this is a digest, not primary data — every figure here ($725B, $55B, IPO valuations, interception rates) needs source verification before it feeds a thesis doc; treat as lead-generation, not evidence.

The biosecurity letter and "answer engine optimization" items are softer but real: the dual-use-virology warning is the kind of frontier-governance signal RDCO watches for downstream regulatory drag on capability, and AEO (gaming what ChatGPT/Google cite) is directly relevant to how RDCO's own surfaces and the Sanity Check newsletter get discovered in an agent-mediated web.

Related