OpenAI and Claude Enter the Consulting Industry Simultaneously

@LiangBinpenny: Not long ago, on May 4th, two things happened simultaneously. OpenAI and Claude both entered the consulting industry.

OpenAI announced the establishment of "The Deployment Company," valued at $10 billion, with over $4 billion in financing. Investors include 19 institutions such as TPG, Brookfield, Advent, and Bain Capital, with SoftBank and Dragoneer also on the list. OpenAI holds a controlling stake and leads operations.

Almost on the same day, Anthropic also announced the formation of an AI-native enterprise services company with Blackstone, Hellman & Friedman, and Goldman Sachs, with committed capital of approximately $1.5 billion. The three main investors each contributed about $300 million, Goldman Sachs invested $150 million, and Apollo, General Atlantic, Leonard Green, GIC, and Sequoia also followed suit.

Two foundational large model companies, on the same day, using almost identical structures, officially marched into the enterprise consulting industry.
This is not a coincidence. This is a declaration of war.
Against whom? The McKinseys of the world.

The global management consulting market is estimated to be around $375 billion in 2025. The MBB trio (McKinsey, BCG, Bain) plus Accenture are the core players in this market. McKinsey's 2024 revenue was about $18.8 billion, but its headcount has already been reduced from 45,000 in 2022 to about 40,000 in mid-2025, with plans to cut another 10% by the end of 2025; BCG's 2025 revenue is $14.4 billion, a 7% year-over-year increase, of which 25%, or about $3.6 billion, comes from AI-related business; Bain's 2025 revenue is about $7 billion.

Interestingly, OpenAI had already signed "Frontier Alliances" with these four in February—McKinsey, BCG, Accenture, and Capgemini are all distribution channel partners for its Frontier platform. But what happened on May 4th is completely different: this is not "selling products through consulting firms," but "becoming a consulting firm yourself," bypassing middlemen to reach customers directly.

What is the essence of consulting? Two words: intelligence + methodology.

Let's talk about intelligence first. The consulting industry's intelligence capability has two layers:
The first layer is public information collection—industry reports, financial data, policies and regulations, market research. Automation in this area has long begun; by 2025, 72% of McKinsey employees are using its internal AI tool Lilli, processing 500,000 queries per month, reportedly saving consultants 30% of research time. BCG's Deckster can automatically generate PPTs, used by 40% of associates weekly. One study shows that such tools can already complete 80% of a junior analyst's work.
The second layer is interviews—the core intelligence source for traditional consulting. Finding industry insiders, suppliers, customers, and former employees for in-depth one-on-one interviews to piece together a "true picture." This is one of the core moats for which consulting firms charge.
But here's the problem: OpenAI and Anthropic have something consulting firms will never have.
Every day, hundreds of millions of users worldwide converse with ChatGPT and Claude. These users are not virtual—they are real enterprise employees, entrepreneurs, investors, engineers, and product managers. They use AI to write business plans, analyze competitors, process supply chain data, research markets, and build financial models. What they submit is not platitudes on a survey, but real problems and real data in real business scenarios.

What is this? This is the largest-scale, unconscious, continuous industry intelligence collection system in human history.
Traditional consulting interviews might conduct 30-50 for a project, at most. Meanwhile, OpenAI's API processes over 15 billion tokens per minute and has 3 million paying enterprise customers. Anthropic has been deployed in the financial services sector at JPMorgan, Goldman Sachs, Citibank, AIG, and Visa. When this data is aggregated, anonymized, and analyzed to form industry insights, its authenticity, timeliness, and granularity are unmatched by any multi-million-dollar interview project conducted by consulting firms.

Interviews yield "what this person is willing to tell you." What users confide in AI is "the real problems they are dealing with." Which is more real?

Now, let's talk about methodology.

McKinsey's "7-Step Method," BCG's Matrix, Bain's NPS—these frameworks are valuable because they distill decades of experience from senior consultants. But essentially, methodology is just workflow: what framework to use to break down what type of problem, what steps to follow, and how to package conclusions into a narrative acceptable to management.
This is precisely what large models excel at distilling.
The "Forward Deployed Engineer" model that OpenAI and Anthropic are now implementing—embedding engineers directly into client companies—is systematically collecting and distilling these methodologies. You embed into a manufacturing company, learn the supply chain optimization process, and after model training, it can be reused for a thousand manufacturing companies. Consulting firms' methodologies are locked in PPTs; large models' methodologies live in parameters. The former requires senior partners to hand-hold junior consultants, while the latter can be infinitely replicated in parallel.
Blackstone's President Jon Gray put it bluntly: this joint venture aims to solve "the biggest bottleneck in enterprise AI implementation"—the scarcity of engineers who can implement cutting-edge AI systems. Goldman Sachs's Marc Nachmann added: "Allowing companies that can't afford consulting fees or recruit AI talent to also access forward-deployed engineer services."
To translate: the core of the consulting industry—people—is being engineered.

What about Palantir?
Some might say there are barriers to entry in areas like military intelligence and national defense. True, Palantir's Forward Deployed Engineer model is exactly the homework OpenAI and Anthropic are copying. But the barriers aren't as high as imagined.
The core reason Palantir has a foothold in the defense sector is security clearances + the trust barrier built by long-term embedding. But technological barriers? OpenAI has secured $10 billion in dedicated deployment funding, and Anthropic has Goldman Sachs and Blackstone behind it—the latter managing over $3.7 trillion in global assets. When your LPs are the world's largest alternative asset managers, you naturally possess distribution channels into hundreds of portfolio companies. This is a distribution advantage Palantir has never had.

Palantir can hold out for a while, but it looks more like using its moat to buy time rather than holding back the tide.

Finally, the conclusion.

The consulting industry used to be technology-intensive and labor-intensive, with very high barriers. Now, when AI models themselves can provide a trinity of "intelligence + methodology + implementation," these barriers don't seem so high anymore.

McKinsey is already laying people off. BCG is desperately packaging AI as a new revenue stream. But what they should really worry about is not AI replacing junior consultants—but AI companies directly replacing consulting firms themselves.

On May 4, 2026, two AI companies crossed industry boundaries simultaneously. This day may be remembered—not because it changed the AI industry, but because it changed the consulting industry.

[Note: The content was co-created by myself and AI]