How CFOs at Chevron, ServiceNow and Nasdaq decide which AI bets create value

CFOs from Chevron, ServiceNow and Nasdaq said AI investments must demonstrate measurable value and compete for capital, while also accounting for the cost of falling behind competitors. ServiceNow expects $500 million in savings this year, largely from using its own products; Nasdaq tracks AI inputs and outcomes including coding speed, client retention and token costs. Chevron says AI supports oil and gas exploration, reservoir recovery and maintenance, and plans to power a Microsoft data center in West Texas through a 2.67-gigawatt natural gas project.

Analysis

The market-relevant signal is not “AI spend” but a shift in underwriting: budgets should migrate from pilot counts toward provable productivity, retention and revenue. That favors vendors able to tie deployment to measurable customer outcomes, while increasing scrutiny on providers whose value case rests mainly on broad productivity claims. ServiceNow’s cited internal savings are a company-reported result, not yet evidence of equivalent customer monetization; watch renewal, net expansion and sales-cycle data for confirmation. Nasdaq’s proprietary data may be a more defensible differentiator than generic model access, but its value depends on clients paying for products built on that data and on manageable data-rights and token costs.

Chevron’s planned gas supply for a Microsoft data center creates potential multi-year power-demand exposure, but the article provides no contract economics, timing or project-status detail. It is therefore not enough to underwrite incremental CVX earnings or infer material MSFT cost relief. Second-order beneficiaries could include gas infrastructure and power-market participants; the corresponding risks are permitting, construction, emissions constraints and customers’ ability to secure power economically.

Contrarian point: “cost of inaction” can justify strategic AI investment before near-term ROI is visible, but it also makes the hurdle easy to game. Investors should distinguish defensible capability-building from activity relabeled as AI. Immediate reaction should be limited; the more useful 1–3 month signals are guidance and disclosed adoption economics, with competitive and power-supply effects playing out over 6–18 months.

Key Decisions for Investors

• No broad AI-beta trade on this panel alone. For NOW, look for reported customer outcomes, renewal/expansion trends and evidence that AI features support pricing; reassess if those metrics fail to improve or sales cycles lengthen.
• Treat NDAQ as a data-monetization watch item, not an automatic AI winner. Verify product revenue/adoption, client retention and token or platform costs; a rising input burden without corresponding client outputs would weaken the thesis.
• Keep CVX’s Microsoft-linked gas project on the catalyst list, but do not capitalize it into earnings estimates until project timing, contracted volumes, pricing and required investment are disclosed. Delays, permitting setbacks or weak economics would falsify the power-demand upside.
• For MSFT, monitor whether reliable power availability becomes a constraint on data-center deployment or a material cost burden. The disclosed Chevron project supports a supply option, but does not establish its scale relative to Microsoft’s overall needs.