Selling the Back Office, Renting It Back

McKinsey eliminated roughly 200 global technology roles this year, part of an effort to automate its own internal functions. Global Managing Partner Bob Sternfels was direct about what that meant.

We will upskill folks more, we will probably have fewer in non-client-deployed areas, but they will be leveraged by today’s technology and AI. — Bob Sternfels

The world's most prestigious consulting firm — the one that sells AI transformation to the Fortune 500 — is hollowing out its own back office under the same banner it sells to clients. ^[1] It is not alone. Visa cut roughly 7% of staff, with CEO Ryan McInerney saying AI was reshaping how work gets done at the company. ^[1] Amazon's Andy Jassy acknowledged that AI would shrink the company's corporate headcount over the next few years. ^[2] Global banks are cutting junior analyst classes by up to two-thirds, with Standard Chartered CEO Bill Winters describing it as replacing lower-value human capital with financial capital. ^[3] By mid-2026, over 142,000 tech-sector jobs had been eliminated, with AI cited in 87,714 US layoffs between January and May alone. ^[4] What looks like a wave of separate cost-cutting decisions is a single mechanism — and it does not stop at the layoff. The firms that cut internal staff citing AI efficiency then encounter a problem: most companies cannot deploy AI at scale on their own. BCG found that only 5% of companies achieve AI value at scale. ^[5] Deloitte puts the share doing deep transformation at 34%. ^[6] ServiceNow's own enterprise index — the vendor's data, not a critic's — shows AI spending rising 110% while operational maturity gained only 16 points, to 51 out of 100. ^[7] The gap between investment and internal capability is wide, and it is the gap the vendors are now moving into. OpenAI has launched a $4 billion Deployment Company that will embed 150 forward-deployed engineers inside client firms. Anthropic followed with Ode with Anthropic, a $1.5 billion venture backed by Blackstone and Goldman Sachs, to do the same. ^[8][9] Blackstone COO Jon Gray made the logic explicit.

I believe it can help break down one of the most significant bottlenecks to enterprise AI adoption by expanding the number of highly skilled implementation partners. — Jon Gray

The bottleneck is the capability gap inside the client. The market opportunity is filling it. TCS, the Indian IT services giant, is positioning itself the same way — cutting 12,000 senior employees while hiring 25,000 graduates — with CEO K Krithivasan arguing that enterprises need outside expertise to scale AI. ^[10] The vendor that replaces the workers becomes the embedded operator of the replacement system. Meanwhile, the workers who remain are losing the skills to operate independently. Researchers describe an "illusion of expertise" in which AI tools boost output while foundational cognitive skills atrophy.

Human cognition is on the obsolescence chopping block. — John Nosta

^[11] And the entry-level pipeline that would rebuild internal capacity is breaking. Entry-level US job postings fell 35% between January 2023 and June 2025. ^[12] Molly Kinder warned of the consequences.

If everyone does that, the entire pipeline of talent starts to collapse and, in a few years, employers in lots of sectors are going to find themselves in trouble. — Molly Kinder

The IMF's managing director gave it a name.

Tasks that are eliminated are usually what entry-level jobs do at present, so young people searching for jobs find it harder to get to a good placement. — Kristalina Georgieva

^[13] Goldman Sachs, meanwhile, has deployed Cognition's Devin agent for production coding and Anthropic's Claude for trades and client onboarding — building dependency on external AI vendors even as it restricts headcount. ^[14] The cycle is not a universal law. IKEA retrained 8,500 call center workers as interior design advisors, generating 1.3 billion euros. ^[15] Some global corporations are using AI at India-based centers to bring functions in-house rather than out. ^[16] Thoma Bravo's Orlando Bravo declined to participate in the AI lab ventures, noting his portfolio companies already use AI tools independently. ^[17] But these exits required deliberate, expensive, multi-year investment in internal capacity — the kind most firms, having already cut the teams that would execute it, have not made. And even companies that try to reverse course find the path blocked. Klarna replaced 850 customer service agents with an AI assistant, then reverted to a hybrid model; 55% of companies that rapidly replaced workers with AI later regretted it. ^[15][18] But rehiring means drawing from a depleted talent pool — the entry-level pipeline is broken — and the workers they retained have been deskilled by the AI they depend on. Reversal routes them back to the same external vendors, because the internal expertise to operate independently no longer exists and cannot be grown back quickly. The cycle does not even save money. Nvidia's Bryan Catanzaro put it bluntly.

For my team, the cost of compute is far beyond the costs of the employees. — Bryan Catanzaro

^[19] Uber exhausted its 2026 AI budget by April. Big Tech announced $740 billion in capital expenditures alongside the workforce reductions. ^[19] Most companies report little to no return on generative AI investments. A Gartner study found that workforce reductions among organizations deploying AI did not correlate with improved financial returns. ^[4] Firms are paying to eliminate their own operational capacity, then paying vendors to replace it, then paying more than they saved. The vendor who engineered the gap is the one you rent from to close it, and the pipeline that would let you stop renting is gone.