A big reset is already underway. Firms are reworking business models as clients demand practical execution over polished answers. Few companies want to shoulder these structural changes alone. Just 17% of technology chiefs in the Goldman Sachs poll expect to develop more software in-house, leaving plenty of room for advisers who can turn plans into results.
"We're seeing the time that people spend on analysis or presentation, relative to the time they spend with clients, shift back to more time with clients," McKinsey CFO Yuval Atsmon told Breakingviews on a recent episode of "The Big View" podcast. He also said the firm is hiring 25% more graduates than usual while keeping team sizes the same.
Meanwhile, AI upheaval is minting demand of its own. Every CEO now faces an avalanche of decisions, from restructuring workflows to budgeting for tokens, the units by which large-language-model usage is metered and billed. Some $400 billion of new business software demand could materialize by 2028, according to Morgan Stanley analysts.
Bleeding-edge labs and cloud computing giants also want in. OpenAI unveiled a consulting venture in May with 19 backers, including investment firms TPG TPG.O and Brookfield, injecting $4 billion. Microsoft started a similar outfit with $2.5 billion and 6,000 staffers, days after Amazon.com pledged $1 billion for its own effort. Anthropic rolled out a $1.5 billion joint venture with private equity firms Blackstone BX.N and Hellman & Friedman, among others, to install its Claude models at midsize companies.
All that firepower pales in comparison to incumbent advice armies and their muscle memories. Accenture counts 85,000 data and AI specialists alone, some 30,000 of them trained on Claude. Little wonder then that OpenAI and Anthropic have both enlisted the firm, alongside Boston Consulting Group, Capgemini CAPP.PA and McKinsey to help move their models from demos into daily corporate use.
At the same time, the older guard is trying to bottle its expertise. Deloitte, led by Joe Ucuzoglu, has spun its smart assistants into a product line backed by a $3 billion investment plan, while EY intends to field 100,000 digital workers by 2028. PwC has already deployed 25,000 people. If it works, firms will be able to sell the same services again and again, lifting profit margins without more hiring in lockstep.
Neutrality also carries benefits. Clients will have reasons to be wary of renting digital brains from a single provider and opt for an agnostic one instead. Roughly 70% of customers at Datadog, whose software monitors large-language systems, already use three or more AI models.
Twelve listed consultancies, worth nearly $300 billion combined, trade at a median cash flow yield of roughly 16%, on 2028 estimates, according to Visible Alpha. At that rate, investors would get their money back in six years. Cheap valuations and costly AI investments create room for consolidation, particularly among firms too small to build the necessary services alone.
Professional partnerships, long reluctant to lean on outside investors, are also raising capital to fund transitions. After years of ignoring calls from would-be buyers, accounting firm Crowe agreed to sell a majority stake to buyout titan KKR KKR.N. The new capital will be used to accelerate AI adoption and fund acquisitions. Size and a strong balance sheet are increasingly the price of admission. Grant Thornton's U.S. tax and audit hub similarly brought in private equity to help unify its international network and acquire additional growth.
At the peak of the advisory pyramid, machines barely register. Boardroom whispering, strategic nous and tactical proficiency trade on relationships, confidence and nerve. A large-language model can produce answers in seconds, but a CEO’s trust will be far harder to earn. Clients will keep paying for human judgment, and for people to blame beyond microchips when things go wrong. To preserve their primacy, however, it's up to consultants to prove they can both dispense and follow quality advice.
ACN•
