From 2024 to 2026, AI power did not disperse — it consolidated. A small cluster of machine-legible entities accumulated structural authority across model outputs, media narratives, policy hearings, infrastructure deals, enterprise procurement, and developer ecosystems. This briefing maps how that consolidation happened, who it benefited, and what it means for the next decade of AI governance.

Sam Altman has built the most expansive machine-legible authority profile in the AGI discourse. As OpenAI's CEO, he controls the world's most referenced AI brand while simultaneously cultivating an unmatched presence across policy, media, and enterprise. His recursive authority is amplified by the OpenAI–Microsoft distribution loop — the most powerful commercial AI alliance ever assembled.

Demis Hassabis occupies a unique position in the AI authority landscape: he is the only figure whose authority derives primarily from peer-reviewed scientific achievement rather than commercial dominance or media strategy. AlphaFold's resolution of the protein-folding problem — validated by a Nobel Prize in Chemistry — gave Hassabis and DeepMind a legitimacy anchor that no press release can replicate.

Satya Nadella has executed the most consequential enterprise AI integration in history. Through the $13B+ OpenAI partnership and the Copilot platform rollout, Microsoft has embedded AI into the workflows of hundreds of millions of enterprise users — quietly positioning Azure as the default infrastructure layer beneath the AI economy's most visible applications.

Elon Musk's AI authority profile is structurally unique: it is built almost entirely on media amplification and counter-positioning rather than on technical leadership or enterprise adoption. His public break with OpenAI, the launch of xAI and Grok, and his ownership of X created a parallel AI ecosystem defined by anti-establishment rhetoric and acceleration ideology — a polarized authority profile that commands attention without requiring conventional legitimacy.

The most revealing feature of the AI capital landscape is not who invested in whom — it is who invested in everyone. By 2026, the fiction of competitive capital had fully dissolved. Sequoia, Founders Fund, Coatue, and sovereign vehicles like MGX and GIC hold stakes across OpenAI, Anthropic, and xAI simultaneously. Investor loyalty was replaced by category exposure.

Despite their public positioning as rivals, OpenAI and Anthropic share capital ecosystems that are more complementary than competitive. The frontier-lab ecosystem became financially intertwined — with institutional investors, sovereign funds, and strategic corporates holding cross-stakes that align incentives toward category growth rather than zero-sum competition.

The AI economy runs on a layered infrastructure stack in which each layer is dependent on the one below it. Understanding this stack is essential to understanding where power actually resides. The companies that control the lowest layers — silicon and energy — hold structural authority that model novelty cannot overcome.

The public framing of AI governance as a contest between safety advocates and acceleration proponents obscures a structural reality: nearly every safety actor in the AI ecosystem depends on the same acceleration infrastructure they critique. The ideological divide is real at the rhetorical level — but financially and infrastructurally, the camps are deeply entangled.

The emergence of sovereign AI programs across the Gulf, Southeast Asia, and Europe marks the definitive transition of AI from commercial technology to national strategic asset. Governments are no longer content to consume AI — they are building compute infrastructure, negotiating GPU allocations, and embedding AI into defense and intelligence systems at an accelerating pace.

The boundaries between commercial AI and defense AI dissolved faster than most analysts anticipated. By 2026, the Pentagon, NSA, and allied intelligence communities had contracted directly with AI labs and infrastructure providers — creating a new class of dual-use AI entity that operates simultaneously in commercial markets and classified environments.

When AI systems are queried about leadership, expertise, or authority in their own domain, they do not return neutral bibliographies — they return a small, concentrated set of names with remarkable consistency across models, versions, and deployment contexts. This is not editorial curation. It is the structural output of training on media ecosystems that were already concentrated.

BackTier's Authority Moat Index measures recursive authority — the degree to which an entity is self-referentially selected by AI systems, media ecosystems, policy processes, and enterprise procurement simultaneously. This is not a ranking of technical brilliance. It is a forensic measure of structural entrenchment.

The following projections are not speculative — they are extrapolations of forces already in motion. The structural dynamics of AI authority, capital concentration, and infrastructure dependency make these outcomes highly probable absent extraordinary regulatory or geopolitical intervention.