CFOs could cut agentic AI costs up to 60% by fixing this overlooked data problem

Companies that prioritize semantics in their AI-ready data will improve agentic AI accuracy by up to 80% and cut costs by up to 60% by 2027, according to new research released at the recent Gartner’s Data & Analytics Summit in London.

The implication for CFOs: a meaningful share of today’s agentic AI spend is at risk of being wasted on tools that hallucinate, introduce bias, and produce unreliable outputs—not because the models are flawed, but because the underlying data lacks context.

“Agentic AI outcomes depend on context, including semantic representations of data,” Rita Sallam, distinguished VP analyst at Gartner, said at the summit. “Without context—a clear understanding of the specific relationships and rules within an organization’s data—AI agents cannot operate accurately.”

Gartner argues that traditional schema-based data models are no longer sufficient, and that a dedicated semantic, or “context,” layer needs to sit at the core of enterprise data infrastructure. Skipping it, Sallam warned, will “perpetuate data inefficiencies” and expose companies to heightened financial, legal, and reputational costs.

For CFOs, the takeaway reframes the AI conversation from a technology debate into a capital-allocation one. Semantic coherence, Gartner says, is becoming “a cost-control and trust strategy, not a nice-to-have,” and potentially a focus for regulators and audit committees evaluating how AI-generated outputs flow into financial reporting and disclosures.

AI remains high on the executive agenda, as data from Q1 earnings calls shows. John Butters, VP and senior earnings analyst at FactSet, recently shared an analysis of S&P 500 earnings calls with me. At least 65% of S&P 500 earnings calls have cited the term “AI” so far, which Butters said slightly below the previous quarter’s 68%. But that prior quarter was the highest percentage going back at least five years, he noted, which makes the current 65% the second-highest share of S&P 500 earnings calls citing “AI” over that period.

When it comes to AI agents, semantics is no longer just semantics. It’s becoming a non-negotiable foundation. Will CFOs need to become linguists, too? Probably not. But they already wear plenty of hats—and the one labeled “chief context officer” may be next on the rack.

Sheryl Estrada
sheryl.estrada@fortune.com

Leaderboard

Manus Costello was promoted to group CFO of Standard Chartered, a multinational banking and financial services company. Costello succeeds Diego De Giorgi, who resigned in February. Standard Chartered had named Peter Burrill as its interim CFO following De Giorgi's departure to join Apollo Global Management as head of the EMEA region. Costello takes on the CFO role after joining Standard Chartered in April 2024 as global head of investor relations. Before that, he spent 25 years in equity research, including as a founding partner and global head of research at Autonomous.

Greg Martini was appointed CFO of Kyverna Therapeutics, Inc. (Nasdaq: KYTX), a late-stage clinical biopharmaceutical company, effective May 18. Martini succeeds Marc Grasso who will be consulting for the company for a transition period. Martini joins Kyverna from Ironwood Pharmaceuticals, where he most recently served as SVP and CFO. Earlier in his career, Martini held finance and corporate development roles at Thermo Fisher Scientific.

Big Deal

Bank of America Institute's latest "Consumer Checkpoint" report finds that spending growth was strong in April. Total credit and debit card spending per household rose 4.8% year-over-year (YoY), up from 4.3% YoY in March, according to BofA internal data. Excluding gasoline, card spending was still a solid 4.0% YoY. However, spending growth slowed in April from March across multiple discretionary "nice-to-have" categories.

Lower-and middle-income households have eased back on discretionary spending, driven by "K-shaped" wage growth.

Higher-income households' after-tax wage growth reached 6 % YoY in April, compared to a more modest 1.5% YoY for lower-income households and 2.3% YoY for middle-income households. That is the widest gap BofA economists have observed between higher-and lower-income earners since the data series began in 2015.

Lower-income households' smaller wage base, combined with slower growth, means their wage gains over the past year have only just covered higher gas spending. 

Going deeper

"NextEra’s $67 billion Dominion takeover creates the world’s largest utility—just in time to win the AI data-center power surge" is a Fortune article by Jordan Blum.

Blum writes: "NextEra Energy’s massive $67 billion deal to acquire Virginia-based Dominion, announced on May 18, will effectively build the world’s largest utility in a bid to dominate the AI data center boom. It’s a goal big enough that NextEra was willing to pay a hefty premium—and risk overpaying—to make it happen. On a call with analysts, NextEra chairman and CEO John Ketchum said the acquisition was necessary to create a player big enough to satisfy enormous and fast-growing demand for electricity." Read more here. 

Overheard

"One of the biggest mistakes companies can make in the age of AI is overlooking early-career talent. Yes, AI is automating some of the more repetitive work that used to define many entry-level roles. However, that shouldn't lead us to pull back on investing in early-career talent. It should push us to rethink how we develop them."

—Gina Mastantuono, president and CFO at ServiceNow, wrote in a LinkedIn post.