AI spending is still accelerating in 2026. Here's how to lead with discipline.
The signal is clear: a new survey shows 68% of CEOs plan to increase AI investment this year. Most projects aren't profitable yet, but leadership teams are pressing forward to avoid falling behind and to keep the story alive for investors.
If you run strategy, you don't need a pep talk-you need a plan. Treat 2026 as a year to scale what works, cut what doesn't, and buy time where the path to ROI is fuzzy.
Why budgets keep growing (even without profits)
- Competitive signaling: Markets reward credible AI roadmaps. Silence reads as drift.
- Moat building: Data collection, fine-tuning, and integration create switching costs later.
- Productivity upside: Early wins in support, coding, and sales ops justify more trials.
- Board pressure: Stakeholders expect visible progress-pipeline, pilots, partnerships.
Source context: An annual CEO survey from advisory firm Teneo points to more spending, not less. See their insights here: Teneo Insights.
Market signal: AI stocks may keep running, but expectations are high
Nvidia has been the prime beneficiary of AI buildouts. With demand for AI chips still intense, the company sits around a $4.6 trillion market cap. Its forward P/E is near 25, above the S&P 500's ~22. That's a premium-and it assumes multi-year spend stays strong.
Even with positive survey data, stocks won't move in a straight line. Nvidia is about 11% off its 52-week high, a reminder that rich valuations can wobble as sentiment shifts.
What this means for your 2026 plan
- Budget with gates: Fund AI in tranches. Each tranche clears a target: accuracy, adoption, cost per task, and margin lift.
- Obsess over unit economics: Track cost per inference, usage by segment, re-prompt rates, and human review time reduced.
- Avoid lock-in: Multi-model procurement (proprietary + open) with swap-out clauses and cost-down schedules.
- Prioritize revenue or COGS: Projects must either add revenue within 12 months or cut unit cost within 6-9 months.
- Production over pilots: Cap pilots at 60 days. If it can't hit KPIs fast, kill it or reroute.
- Data readiness first: Clean, well-permissioned data beats bigger models. Invest there before more experiments.
- Security and compliance built-in: Model governance, audit trails, PII controls, and vendor risk reviews as default.
Operator and investor takeaways (keep it simple)
- Operator lens: Treat AI like infrastructure. De-risk with clear TCO models, strong vendor terms, and real adoption metrics.
- Investor lens: Favor cash-generating "picks and shovels" and platforms with line-of-sight to demand. Be wary of application names priced for perfection.
Executive checklist to keep AI spend honest
- Do we have a per-use case ROI model tied to margin, churn, or cycle time?
- What is our cost per interaction, and how will it decline over 12 months?
- Where does AI remove headcount hours versus assist them? Quantify both.
- Is accuracy above a business-safe threshold? What's the rollback plan?
- How often do we retrain or update prompts, and who owns that process?
- What's our vendor exit plan if pricing spikes or performance dips?
- How are we measuring user adoption and retention beyond vanity metrics?
- Which projects move the P&L this year, not in three?
- What controls protect IP and customer data across vendors and models?
- Which wins are ready to scale globally, and what's the sequencing?
Where to invest, where to pause
- Invest: Customer support deflection, sales enablement, developer productivity, internal search, and pricing/forecasting models with clear payback.
- Pause: Flashy demos without owners, one-off copilots that don't touch core KPIs, or projects with murky data rights.
If you're upskilling teams
Point managers to practical programs that connect tools to outcomes and role-specific KPIs. A focused path by role helps you turn pilots into production faster.
The bottom line: AI budgets will keep rising in 2026, but returns won't come from spend alone. Returns come from ruthless prioritization, clean data, clear KPIs, and contracts that let you scale up-or walk away-on your terms.
Your membership also unlocks:
