Billionaire VC Chamath Palihapitiya has fired a blunt warning at the consulting industry: by embedding OpenAI and Anthropic models directly into their client workflows, firms like PwC and Accenture are effectively “letting the fox into the hen house.”
Palihapitiya posted the criticism on X on May 17, calling out consultancies that route client work through LLM providers. “OpenAI and Anthropic are openly funding and starting competitors to you while also using your usage to drive more success for them,” he wrote, arguing this isn’t just risky — it’s structurally self-defeating.
Why Palihapitiya is alarmed
- OpenAI and Anthropic have moved beyond selling models to building enterprise services that compete with traditional consultancies. OpenAI recently launched the OpenAI Deployment Company, backed by more than $4 billion from 19 institutional investors at a $10 billion pre-money valuation. Investors include Goldman Sachs, TPG, Brookfield, Bain Capital, McKinsey & Company, SoftBank and Capgemini.
- Anthropic followed with its own enterprise services arm, supported by $1.5 billion from the likes of Blackstone, Hellman & Friedman and Goldman Sachs.
- Combined, the two labs have committed roughly $5.5 billion to build consulting and implementation capabilities — positioning them to displace the very firms that help roll out their models.
The economics are already shifting
- Anthropic’s run-rate revenue reportedly reached $30 billion in early April 2026, up from $9 billion at the end of 2025, with more than 1,000 enterprise customers each spending north of $1 million per year.
- OpenAI crossed an annualized $25 billion revenue run-rate as of February 2026, with enterprise contracts accounting for over 40% of total revenue.
- Both labs are hiring engineers and embedding forward-deployed teams inside client organizations — a playbook that historically belonged to consultancies like Accenture and PwC.
Palihapitiya’s alternative: own the control plane
Palihapitiya argues consultancies should stop being simple resellers of LLM compute and instead build model-agnostic “control planes” that arbitrate which data and tokens go to which provider. “Controlling the tokens is controlling the spice,” he wrote, borrowing from Dune.
He pointed to his own startup, 8090, as an example. In March 2026 8090 announced a global partnership with EY: EY adopted 8090’s Software Factory platform to power EY.ai PDLC, an AI-native product development lifecycle tool that EY says will boost developer productivity by 70% and accelerate delivery by 80x. The platform is being rolled out to tens of thousands of EY US consultants. Palihapitiya says 8090 can route token generation to any model provider and is close to announcing another global partnership.
What this means for consultancies — and crypto players
Palihapitiya’s warning is both practical and strategic: if consultancies continue piping client workflows straight through OpenAI or Anthropic, they risk training and empowering their own competitors. The labs’ move into implementation and embedded engineering teams is already replicating — and potentially replacing — traditional consulting services.
For crypto and Web3 audiences, the debate has an obvious resonance: control over “tokens” (in AI usage or in crypto) determines who captures value and who can build downstream products. Building model-agnostic middleware — analogous to protocol-agnostic infrastructure in crypto — could be the difference between staying relevant and being displaced.
Palihapitiya’s closing barb captured the stakes: consultancies that “refuse to accept the disruption” or worse, accelerate it by adopting the disruptors’ tools, are setting themselves up for obsolescence. The window to build independent, model-agnostic infrastructure, he warns, may be closing faster than many realize.
Read more AI-generated news on: undefined/news
