AI Token Volume Projected to Surge 78% as Lower Unit Rates Fail to Contain Enterprise Budgets: Accenture

Corporate technology executives banking on falling model pricing to rein in software expenditures are facing a fiscal reality check. A new report from Accenture Research reveals that enterprise spending on artificial intelligence will continue to rise over the next two years, driven by an explosion in consumption that easily outpaces declining per-unit rates.

The analysis projects that while average token prices will drop roughly 19% over the next 24 months, corporate token consumption is expected to surge by 78%. This divergence triggers a widespread computational rebound effect: cheaper access is prompting businesses to expand deployments across more departments and launch compute-heavy tasks that were previously shelved.

The Rebound Mechanism: Why Savings Won't Hit the Bottom Line The study illustrates that corporate IT departments rarely pocket the dividend from falling model costs:

• Negligible Cash Retention: If inference rates drop by 25% or more, only 5% of surveyed organizations plan to bank the savings directly to their bottom lines.

• Expanding the Seat Count: Instead, 42% of firms intend to roll out AI tools to wider swaths of their employee base, while 35% plan to deploy complex applications that were historically deemed cost-prohibitive.

• Volume Outstripping Discounts: Because lower transaction barriers accelerate experimentation and continuous agentic loops, net IT outlays across AI infrastructure are climbing rather than contracting.

"Lower unit costs do not necessarily translate into lower overall spend. Businesses must evaluate the total cost of AI-enabled work rather than focus only on falling prices." — Accenture Research

The Outcome Calculation Deficit Despite board-level scrutiny on software returns, most organizations are scaling model utilization without basic economic visibility:

The ROI Blind Spot: Just 35% of companies can accurately determine the cost per business outcome for their highest-volume AI deployment, leaving the majority unable to gauge actual financial returns.

• Unmonitored Operating Overhead: Without granular, workload-level tracking, organizations risk converting AI from a calculated capital asset into an unchecked, highly volatile operational expense line.

Augmenting Talent Over Mass Displacement The report counters the narrative of wholesale workforce reduction, indicating that enterprises are predominantly using artificial intelligence to amplify existing staff bandwidth:

• Workforce Augmentation: A substantial 70% of organizations confirm AI is primarily deployed to augment human workers, compared to just 4% using it principally to eliminate headcount.

• Reinvesting Reclaimed Hours: C-suite leaders are actively planning how to redirect freed-up employee capacity, with 60% intending to funnel saved time into high-value strategic work and 51% planning to channel employee focus into revenue-generating growth initiatives.

Accenture cautions CIOs that successful AI governance will depend less on hunting for marginal price cuts from model providers and far more on establishing strict outcome attribution—ensuring every token processed maps directly to measurable enterprise value.