BCG: Companies plan to double AI spend in 2026 - what executives should do now
AI has moved from pilot to priority. A new report shows companies plan to double AI investment in 2026, targeting roughly 1.7% of revenues. That's not a side bet. It's a line item tied to growth, margin, and competitive position.
The signal is clear: top leadership is taking ownership. CEOs are stepping in as the main decision makers on AI, reallocating budgets beyond IT to fund skills, data, and agentic systems built for measurable outcomes.
What's changing at the top
- Ownership: Nearly three quarters of CEOs now say they are the main decision makers on AI.
- Accountability: About half believe their jobs depend on AI outcomes. Four out of five are more optimistic about ROI than they were a year ago.
- Time investment: Leading CEOs are spending 8+ hours per week on their own AI upskilling and are investing 2x more than peers in capability-building across their orgs.
As BCG's leadership notes, AI is no longer a side project for IT or innovation teams. It's influencing strategy and operations from the top, and boards are watching.
Where the money is going
- Budget levels: Companies plan to reach ~1.7% of revenue on AI in 2026, doubling current spend.
- Agentic AI: More than 30% of this year's AI investment is going to agentic systems (autonomous or semi-autonomous agents that perform tasks and workflows).
- Conviction: ~90% of CEOs believe AI agents will deliver measurable ROI in 2026.
- Durability: 94% of CEOs say they'll maintain or increase AI investment even if near-term returns lag.
The bet is shifting from "tools and trials" to "systems and outcomes." Agentic AI is being positioned to own processes, not just assist with tasks.
China's stance: stronger conviction, faster moves
- 94% of Chinese CEOs say they will increase AI investment even without immediate returns.
- 89% are more optimistic about AI ROI than last year.
- 91% say they lead AI strategy, and ~60% believe their career stability depends on its success.
This mindset-commitment ahead of returns-often precedes share shifts. If your competitors hold that posture, waiting is the risky move.
What this means for your 2026 operating plan
- Set a real number: Anchor your AI budget against revenue. If you're far below ~1.7% by 2026, you're likely underinvesting.
- Fund beyond IT: Allocate meaningful budget to data quality, change management, process redesign, and upskilling. The ROI lives there.
- Make the CEO the owner: Centralize AI decisions at the top with a small steering group (CFO, COO, CHRO, CIO/CDO). Monthly reviews. Clear gates.
- Pick three needle-movers: Prioritize agentic AI for revenue, cost, and cycle-time. Examples: sales ops agents, finance close agents, customer service case resolution agents.
- Tie pilots to P&L: Each pilot needs a baseline, target KPI, timebox, and exit criteria. No indefinite experiments.
- Upskill leadership first: Require an 8-hour weekly block for executive AI education for the next 90 days. Leaders model the behavior the org will follow.
- Governance without friction: Stand up lightweight guardrails for data, model usage, and approvals. Fast lanes for low-risk use cases; escalations for high-risk.
- Vendor and model strategy: Avoid single-threading. Use a portfolio of providers and models matched to use cases, cost, risk, and latency needs.
A simple sequencing plan
- Q1: Executive education; pick top three use cases; secure data access; define KPIs; set governance.
- Q2: Build and pilot agentic workflows; integrate with core systems; track weekly ROI and adoption.
- Q3: Scale winning pilots; codify processes; train frontline teams; renegotiate vendor contracts.
- Q4: Expand to adjacent functions; refine cost models; lock in annualized benefits and opex forecasts.
Metrics that matter
- Revenue: Qualified pipeline lift, win-rate lift, cross-sell rate.
- Cost: Cost per ticket, cost per close, cost per lead, model inference costs.
- Speed: Cycle time to proposal, days to close, time-to-resolution.
- Quality and risk: Error rates, audit findings, policy exceptions.
- Adoption: Weekly active users, task coverage by agents, human-in-the-loop acceptance rate.
Why agentic AI is getting budget share
Executives expect returns where software runs the work, not just assists it. Agents can plan, take actions across tools, and report back with outcomes. That's why they're earning 30%+ of AI budgets this year.
For a concise overview of enterprise AI and agentic systems, see BCG's AI perspective here.
Avoid the common failure modes
- Tool-first thinking: Lead with business problems and KPIs, then pick models and platforms.
- Pilot sprawl: Limit to a portfolio you can actually measure and staff. Kill weak pilots fast.
- Data neglect: Poor data quality erases ROI. Budget for cleanup and access early.
- Change fatigue: Treat enablement as part of the product. Training, comms, incentives.
- No owner: If everyone owns it, no one owns it. Put a name next to each outcome.
If you're building executive-level skills
Block time for focused upskilling now. Start with decision frameworks, governance, and high-ROI use cases your teams can execute in 90 days.
Bottom line
Budgets are doubling. CEOs are in the driver's seat. Agentic AI is where a big share of spend is going because that's where measurable ROI is expected in 2026.
Set a target, pick a few high-leverage use cases, and commit leadership time. The companies that treat AI as an operating priority-funded, measured, and owned-will set the pace next year.
Your membership also unlocks:
