Tiffany Luck, a partner at venture firm NEA, said enterprises still cannot clearly grasp or measure the return on their AI investments, noting that general-purpose models alone rarely translate into concrete business outcomes.
Enterprise AI · The ROI Reckoning
Enterprises Are Still Figuring Out Their AI ROI
As the "tokenmaxxing" boom gives way to a cost reckoning, only a small slice of companies report significant, measurable returns — pushing focus toward vertical AI that owns the full workflow and delivers a finished work product.
56%
of CEOs report neither revenue gains nor cost cuts
84%
expect positive returns to take over 6 months
53%
of companies see only 1–5% ROI
The signal vs. the noise
Who actually reaches significant, measurable ROI?
~10%
achieve significant ROI
~90%
still chasing measurable returns
Each block ≈ 10% of agentic-AI adopters. Only about 1 in 10 has realized significant returns.
The "last mile" gap
Horizontal models cover only 0–80% of the work — the remaining last mile breaks ROI.
Re-forecasting context
Flagging trade-offs
Integrating data layers
Data provenance & audit
WHAT WORKS
Vertical AI owning the full workflow end-to-end
Delivering a finished work product — reports, due-diligence docs
Forward-deployed engineers embedded with customers to build a moat
Smaller, flexible firms transforming processes faster
WHAT STALLS
Projects stuck in the pilot stage
Friction integrating into real operations
Budget overruns from the tokenmaxxing trend
No measurable financial return within six months
The new yardstick
The path ahead is rethinking evaluation around hours saved , error reduction , and quality improvement — moving AI from a shiny object to real, measurable integration.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…