AI Adoption Outpaces EBIT Gains, TokenPost Says

August survey finds wide use, limited EBIT impact

TokenPost reported on October 1, 2026, that an August 2026 survey found 89% of organizations regularly used AI in at least one business function. John Kim, who reports on the digital-asset business for TokenPost, ran the findings under a headline saying AI adoption outpaces companies' reported financial gains.

Among respondents, 37% attributed some positive impact on earnings before interest and taxes (EBIT) to AI, TokenPost said. Only 6% qualified as high performers by citing significant impact and attributing at least 5% of EBIT to AI, according to TokenPost. The survey included 1,719 respondents and ran from May 4-June 8, 2026, the outlet said.

Kim wrote that the findings show a gap between regular use and the impact respondents described, and he cautioned that the figures reflect survey responses. The results highlight the challenge of turning task-level productivity improvements into broader company results, according to TokenPost.

Leger index finds training and returns lag use

Leger360 published Leger's 2026 AI Index on October 1, 2026, finding that 68% of Canadian employees use AI in their day-to-day work while 61% have received no AI training at work. Among organizations using AI, 46% report no measurable benefit, 21% only marginal benefit, 30% operational benefit and 3% meaningful business results tied to revenue or ROI, Leger said.

The business component of the index drew on a survey of 3,031 workers and 1,050 AI decision-makers, according to Leger. Only 27% of employees consider themselves highly skilled at using AI effectively at work, and 17% of organizations using AI say they provide structured support, the index found.

Leger also found that nearly one in two employees sometimes uses an unapproved or unofficial AI tool for a work task. "A lack of measured impact does not necessarily mean AI is creating no value," Leger wrote in its release.

BearingPoint finds scaling stalls past pilots

Only 13% of companies were on track with their AI initiatives, consultancy BearingPoint found in a study covered by Reuters and carried by Investing on October 1, 2026. Regulatory hurdles and integration with legacy IT systems slowed large-scale adoption, the consultancy said.

Frederic Gigant, a BearingPoint expert, said "AI has crossed an important threshold" and added that proving value and scaling it were two different things, Globalbankingandfinance reported in its October 1, 2026, version of the Reuters story. Nearly three-quarters of surveyed companies reported positive financial results from AI, but less than a third could move beyond pilot projects, BearingPoint said.

About 40% of respondents cited legal regulations as the main barrier to scaling AI, and 34% pointed to integrating the technology into existing IT systems, Reuters said. Around 24% of companies reported AI-driven cost savings of at least 10%, against 4% reporting revenue growth of the same scale, BearingPoint found.

Among surveyed companies in the United States, 18% reported comprehensive AI implementation, compared with 20% in China and 8% in Germany, BearingPoint found. The share of companies with AI deeply integrated into operations rose to 11% in 2026 from 7% in 2025, the study showed.

Task-level studies measure narrow settings

TokenPost also cited task-level research in which AI assistance increased issues resolved per hour by an average of 15% in a study of 5,179 customer-support agents at one company. The research measured different tasks and groups of workers and does not establish that comparable improvements will transfer to other business processes, TokenPost cautioned.

A separate survey published in July found that 32% of leaders at organizations that redesigned workflows reported capturing enterprise value, compared with 6% of leaders at organizations that did not, according to TokenPost. The July survey covered 750 employees and leaders, and the comparison drew on a subset of 582 leaders, the outlet said. TokenPost wrote that the responses show an association between workflow redesign and value capture but do not establish that redesign alone caused the difference.

Hotel chains and Bank of America post results

In hospitality, an h2c industry study released October 1, 2026, found that 91% of participating hotel chains already use AI, Hotelnewsresource reported. Only 28% of respondents said their company has an organization-wide AI strategy directed by senior leadership, and 13% of hotel chains cited measurable returns from AI, according to h2c.

Nearly 70% of respondents reported efficiency and automation improvements, while 56% named a lack of expertise, skills or training as a constraint, h2c found. The study drew 122 responses representing 113 hotel chains and was not weighted by chain size, Hotelnewsresource said.

At Bank of America, chief technology and information officer Hari Gopalkrishnan oversees a roughly $14 billion annual technology budget that includes more than $4 billion for AI and other new initiatives, Fortune reported on September 29, 2026. Erica for Employees is used by 90% of the bank's workers, and the EricaAssist agent, deployed to more than 18,000 customer service representatives, helps reduce average call times by nearly one minute per interaction, Fortune said. "Pretty much everything we demoed a year ago is now live in a scaled fashion," Gopalkrishnan said, according to Fortune.

Tron's take

My take: these surveys describe a measurement problem more than an access problem. Most organizations in the TokenPost and h2c samples already use AI, and most Canadian employees in Leger's sample do, yet far fewer respondents can tie AI to revenue, EBIT or measurable returns. That is my reading of the news, not a reported result. The samples also differ: Leger surveyed Canadian workers and AI decision-makers, h2c surveyed hotel chains, and every survey figure here is self-described by respondents.

I disagree with two common positions. One says a small business must adopt every new AI release immediately to stay competitive. The high performers in these surveys are rare, and Gopalkrishnan describes Bank of America's progress as tools demoed a year earlier now running at scale, not this week's releases. The other position says AI news is irrelevant to small firms. The regulation and integration barriers BearingPoint found, and Leger's finding on unapproved tools, apply to small offices as much as to banks, so owners benefit from knowing what shipped and what stalled.

My advice: pick one or two processes, record a baseline before rollout, and expand only where the measured number moves. I would also inventory the AI tools staff already use before buying more. XL.net sells managed IT services and security assessments, including that kind of tool inventory, so weigh my advice with that in mind. Earlier XL.net coverage of CompTIA's workforce readiness findings and Salesforce's agentic AI ROI pitch covers the training and vendor sides of the same question.