Williams Companies Stock Soars on AI Power Demand. Is It Too Late to Get In?

Williams reported Q3 adjusted EPS of $0.54 versus $0.5386 consensus and revenue of $3.2 billion, 10.1% above expectations; adjusted EBITDA rose 6% year over year to $2.04 billion. The company expects to reach the upper half of its 2026 adjusted EBITDA guidance of $8.3–$8.5 billion and reaffirmed a target of at least 11% annual EBITDA growth through 2030, supported by AI-related power projects and its $2.82 billion Momentum Midstream acquisition. Shares rose 2.7% to $73.09 after results, but leverage increased to 3.80x and 2026 growth capex is forecast at $7.3–$7.9 billion; the 2026 dividend was raised 5% to an annualized $2.10.

Analysis

The key question is not whether AI raises U.S. power demand, but how much of that demand Williams can convert into contracted, incremental cash flow after construction costs. Behind-the-meter gas supply could bypass slow grid interconnections and support utilization of Williams’ network; it may also benefit gas producers and gas-fired generation suppliers. But data-center load forecasts are not equivalent to Williams-addressable volumes, and onsite projects may require substantial capital before producing returns. The economics and contract terms for Socrates, Plato North and future projects remain the critical missing evidence.

Near term, the earnings update supports the growth narrative but does not establish an earnings inflection: EBITDA growth was 6%, while the 11%+ through-2030 target and upper-half 2026 outlook are company guidance. With a premium valuation, the bar is for project execution and returns—not simply additional announcements. Over 1–3 months, commissioning and disclosure of commercial terms are catalysts; over 6–18 months, capex conversion, leverage and realized EBITDA will determine whether the premium holds. Higher construction costs, permitting slippage or rates could pressure returns and the multiple. A contrarian risk is that investors capitalize broad AI electricity forecasts into WMB before proving that specific projects generate attractive incremental cash flow. The thesis weakens if projects slip, guidance is reduced, or leverage rises without corresponding EBITDA growth.

Key Decisions for Investors

• Do not chase the post-results move on the AI-demand narrative alone. Keep WMB on a watchlist and add only on weakness or after the company provides verifiable project economics: contracted capacity, expected in-service dates, capital required and incremental EBITDA/cash flow.
• For a relative-value expression, consider a modest long WMB against a diversified utilities exposure only if new project disclosures confirm contracted demand and returns. The intended edge is faster monetization of constrained power demand; the main risk is that data-center demand is delayed or WMB’s project returns do not justify its valuation premium. Avoid sizing this as a pure AI-demand trade.
• Track quarterly EBITDA growth against the 11%+ long-term target, growth-capex execution, leverage and project commissioning. Falsify the constructive thesis if guidance is cut, key project dates slip, or leverage continues rising without demonstrated EBITDA contribution.
• Treat the dividend as support, not a valuation floor: higher rates and capital needs can still compress the multiple even with stated dividend coverage. Reassess if financing costs rise materially or coverage deteriorates.