IBM: AI Accounts for 30% of Consulting Revenue, Margin Shrinks

IBM reported on Wednesday (21) that work related to artificial intelligence accounted for about 30% of its consulting division's revenue in the third quarter of 2026 and 35% of its contract portfolio, according to CFO Jim Kavanaugh during a conference call with analysts. In the same quarter, the consulting gross margin fell to 27.4%, down from 29.3% a year earlier and 28.9% in the second quarter. Together, these two pieces of information are the most useful data from the balance sheet for those selling technology services.

The consolidated results came in above expectations. Revenue totaled $17.1 billion, up 5% (6% in constant currency), against a consensus of $16.82 billion, while adjusted earnings per share were $2.95, compared to the projected $2.86. Nonetheless, shares dipped in after-market trading, as reported by CNBC and Investing.com, with the market focused on the margin.

Software Drives, Consulting Stalls

The software division generated $7.8 billion, reflecting an 8% increase (9% in constant currency), and served as the main driver for the quarter. Consulting remained stable at $5.3 billion, with a 2% increase when excluding the currency effect. For 2026, IBM expects low to mid-single-digit growth in consulting and has maintained its projection of 4% to 5% revenue expansion in constant currency. The free cash flow target has increased: the company now anticipates growth exceeding $1 billion over 2025.

Part of the strength in software came from contracts that had been delayed but closed in this quarter, according to summaries from the conference call. This caveat matters: a 9% growth driven by backlogged business does not guarantee the same pace in the fourth quarter.

Shrinking Margins by Choice

Kavanaugh attributed the margin decline to a deliberate decision: IBM is reinvesting productivity gains into digital assets as it transitions to what it calls a "services-as-software" model. Practically, the company utilizes the efficiency gained from AI in its delivery teams to build reusable agents and platforms, rather than returning that efficiency to the bottom line.

The skeptical reading is straightforward. If AI makes each consultant more productive and the client starts paying for results, project revenue is likely to decline before the new assets generate recurring revenue. The gross margin of 27.4% is the first quantitative evidence of this gap. The optimistic reading also has concrete data: a portfolio where 35% of contracts already involve AI indicates that clients are purchasing the service, not just testing it.

The distinction lacking in the debate is between volume and mix. IBM’s consulting revenue did not grow in dollars, but the share of AI in revenues reached 30% and 35% in the contract portfolio. The five-point difference indicates that the AI share should grow in the coming quarters; the question is whether it will replace traditional projects with higher margins or supplement them. The balance sheet does not separate the two effects.

Three Markets, Same Pressure

In the United States, IBM's results raise the question that analysts will pose to Accenture, Cognizant, and other integrators: How much of the productivity generated by AI remains with the company, and how much is passed to the client in pricing? A consulting firm whose revenue remains stable while its gross margin falls by 1.9 points over a year is, in practice, financing its own transition through today’s results.

In India, where IBM and its competitors concentrate a significant portion of their delivery workforce, the "services-as-software" model alters what is contracted. Fewer hours of development and testing, more engineers to build and operate agents. If 35% of the portfolio is already AI-related, the composition of the Indian teams hired for these projects shifts before revenue, with the effect appearing first in hiring and only later in revenue.

In Brazil, IBM Consulting competes for contracts with banks and the public sector against Accenture, the Big 4, and local integrators. A Brazilian client reviewing IBM’s 27.4% margin has a new argument on the negotiation table: if the supplier is already gaining productivity with AI, the price for delivery should reflect that. The same argument applies to a German bank or a Japanese insurer.

The point left open by the balance sheet is speed. IBM has demonstrated that AI is already a significant part of what it sells in consulting and that the margin finances the transition. It has not shown when the digital assets it is reinvesting in will start appearing as software revenue, and that date—not the percentage of AI—will determine whether the model works.