Good morning. AI is shifting quick, however many corporations nonetheless haven’t determined who ought to personal the job of turning that momentum into measurable enterprise worth.
At Fortune’s Fashionable CFO dinner in San Francisco final Thursday, sponsored by Deloitte and ServiceNow, Melissa Valentine, a senior fellow at Stanford Institute for Human-Centered Synthetic Intelligence, delivered a transparent message to CFOs: they’ve a narrowing window to take command of AI worth creation.
Valentine pointed to a latest Harvard Enterprise Evaluation article by the founders of the Return on AI Institute, citing survey findings that underscore this opening. Solely 2% of the C-suite executives surveyed mentioned CFOs have been charged with capturing worth from AI. But when CFOs have been accountable, 76% reported producing substantial worth, properly forward of different capabilities. Laks Srinivasan, coauthor of that report, informed me that finance chiefs are uniquely positioned to outline, consider, fund, and measure AI initiatives, then apply that framework throughout the corporate.
Valentine, a tenured affiliate professor of administration science and engineering at Stanford’s College of Engineering, informed the room of finance chiefs that CFOs have a strategic opening to steer on AI if they’re keen to quantify the worth and be accountable for it. She argued that generative AI is shifting out of its experimental section and into one thing CFOs know properly: systematic measurement. Two years in the past, she mentioned, rigorous accountability would have been untimely. At the moment, it’s important.
On the query of guardrails, Valentine pointed to a latest incident during which Anthropic inadvertently uncovered inside supply code for its Claude coding instrument, providing a uncommon public glimpse into how frontier AI labs defend their fashions. She referred to as consideration to the idea of “harness engineering,” the infrastructure surrounding fashions to make them usable and secure, together with secondary AI techniques designed to observe main ones. Her recommendation to CFOs: Research that structure as a result of leaders should perceive whether or not the system round a mannequin is powerful sufficient to control, monitor, and belief at enterprise scale.
That instance bolstered a broader level in Valentine’s remarks: the necessities for secure, production-grade AI are basically completely different from these for on a regular basis worker experimentation. She drew a pointy distinction between two very completely different types of AI transformation. One begins on the frontline, the place workers use instruments resembling Gemini or NotebookLM and uncover sensible functions via experimentation. The opposite is pushed from the highest, the place production-grade use circumstances demand sturdy information infrastructure, engineering rigor, and governance. Each matter. Every requires a definite working mannequin.
Leaderboard
Marcel Teunissen was appointed CFO of Increase Power Company (Nasdaq: EXE), efficient April 6. Teunissen most lately served as president of North America for Parkland Company. He beforehand served as Parkland’s CFO the place he led the corporate’s monetary technique, capital markets, and investor engagement. Earlier than Parkland, Teunissen spent greater than 20 years with Shell plc in roles, together with as VP of finance for Built-in Fuel Ventures and EVP of finance for International Built-in Fuel and New Energies.
Steven E. Pfanstiel will step down from his function of EVP, CFO and treasurer of Neuronetics, Inc. (Nasdaq: STIM), a medical expertise firm. Pfanstiel is pursuing a possibility exterior the corporate. He’ll stay via Might 1. Neuronetics has launched a search to determine his successor.
Large Deal
As corporations construct expertise pipelines for the subsequent technology of workers, AI is shaping how faculty college students take into consideration their educational paths, based on the findings of a Lumina Basis-Gallup 2026 State of Increased Training Research. Forty-two % of bachelor’s diploma college students surveyed say AI has led them to think about altering their main, together with 13% who’ve considered it a fantastic deal. Amongst affiliate diploma college students, that share rises to 56%, with 15% giving it critical consideration.
One other key discovering is that whereas conventional motivations — together with gaining expertise, increased pay and profession achievement — stay way more widespread, about one in seven bachelor’s diploma college students (14%) and affiliate diploma college students (13%) say making ready for AI and different technological advances is a vital motive they enrolled.
Going deeper
The State of AI in Manufacturing is a report by Digit, primarily based on an evaluation of U.S. Census Bureau Enterprise Tendencies and Outlook Survey, information from manufacturing corporations from July 2022 to February 2026. AI adoption in U.S. manufacturing has grown 4 instances since 2023. Enterprises are additionally 2.3 instances extra more likely to undertake than small producers.
However that development is gradual and cautious — 87% of producers nonetheless have not adopted it. Uncertainty round adopting AI has grown from 9.2% to 14.4%. An element contributing to gradual adoption is the dearth of readability about AI’s ROI.
What producers are lacking is a transparent drawback to unravel to maneuver from planning to execution, based on the report.
Overheard
“Most organizations trying to deploy AI are discovering that the hardest problems are not technological. Data readiness, security, integrations, workflow redesign, and building human skills remain stubborn bottlenecks for true AI implementation.”
—Omar Abbosh, CEO of Pearson, a worldwide training firm, writes in a Fortune opinion piece.
