Dr. Jonah Tebaa has been tracking a pattern across the applied-AI projects he advises on in Lebanon and the wider MENA region: the technology almost never fails. The finance function does — or rather, the mismatch between how AI vendors bill and how regional companies move and budget money does. In his recent writing, he traces this back to a specific commercial-design problem that most AI vendors, analysts, and even AI consultants working in the region rarely name directly.
A gap in the commercial model, not the model
Dr. Tebaa's argument starts from an observation he has made repeatedly in client engagements: AI pilots in the region routinely clear their technical bar. Forecast accuracy improves, automation targets get hit, stakeholders sign off. What stalls the rollout is what happens next — the first full invoice, priced in USD and scaled to actual usage, landing on a finance team's desk built around a fixed annual budget and a banking system that does not move dollars the way vendor contracts assume it will.
He identifies three friction points that, in his view, are almost never surfaced during vendor evaluation. The first is currency and banking rails — most AI vendors offer no local-currency billing, leaving companies exposed to exchange-rate movement and correspondent-banking delays on every dollar transfer. The second is the structural conflict between consumption-based pricing, which has no ceiling by default, and the fixed annual budget line items that dominate procurement culture across the region. The third is procurement processes themselves, which he notes were largely designed for one-time software purchases and rarely include a review mechanism for a monthly AI relationship that evolves in scope and price over time.
Why the standard playbook doesn't transfer
What distinguishes his framing from more generic commentary on "AI readiness" in emerging markets is his insistence that this is not a maturity problem or a change-management problem — it is a design mismatch. Global vendors build pricing models for economies with stable currencies and flexible monthly spend. Analysts and consultants trained on those markets tend to inherit the same blind spot, evaluating AI deals on technical fit and strategic upside without pressure-testing the commercial terms against local banking and budgeting reality. Dr. Tebaa's position is that this gap gets discovered by finance teams after signature, when it is far more expensive to renegotiate than it would have been to raise before the contract existed.
A negotiable checklist, not a warning
Rather than framing this as a reason to delay AI adoption, Dr. Tebaa treats it as a solvable procurement problem. He advises clients to secure specific, written terms before signing any applied-AI contract: a local-currency or fixed-USD-equivalent billing option, a confirmed and tested banking rail for payment, a hard usage cap or spend alert set by the client rather than discovered on an invoice, an annualized or committed-spend structure that maps onto fixed-budget approval cycles, and a recurring contract review clause tied to actual usage data. He also flags the internal decision of framing the AI line item as opex versus capex as something companies should resolve before the vendor conversation begins, since it determines who has authority to approve overages later.
The throughline in his thinking is that finance and procurement should be brought into an AI rollout at the start, not treated as a formality once the pilot succeeds. In his framing, the executives who get AI adoption right in Lebanon and MENA are the ones who involve their CFO in shaping the commercial terms of the deal from the first vendor conversation, rather than presenting a signed contract for approval after the fact. It is a distinction he returns to often: in a region where currency and banking friction are operating realities, treating the finance function as a strategic partner in AI adoption — rather than a downstream approval step — is what separates projects that scale from projects that quietly stall.