NEW RESEARCHThe Next AI Market Is Built Before It Is SoldRead now ↗
EXECUTIVE DECISION BRIEF · 01 SEP / 2026DAILY EDITION · 7 MIN READ

THE AI INSTITUTE / RESEARCH FOR LEADERS

The Next AI Market Is Built Before It Is Sold

Model access does not create demand. Identity, payments, connectivity, local context and institutions decide whether a service becomes a market.

OUR VIEW

A viable AI market requires five complements around the product: anchor demand, delivery rails, local context, operating capability and workable rules.

Key points

What this paper means for leaders

  1. Estimate reachable demand from actual access and use—not population or network coverage alone.
  2. Secure one anchor customer or public-service workflow before committing large distribution or infrastructure capital.
  3. Fund the identity, payment, connectivity and data rails the service needs to complete a transaction or outcome.
  4. Adapt language, workflow, support and pricing through institutions that remain in the market.
  5. Stop or sequence entry when the complements cost more than the service can plausibly return.
01

The market signal

The useful part of GITEX Nigeria is what sits beside AI

GITEX Nigeria's programme spans Abuja and Lagos from 31 August to 3 September. Its organiser says the event connects more than 150 federal and state agencies, 40 banking institutions, 300 investors and 750 startups. Those figures establish the breadth of a convened ecosystem. They do not establish adoption, deals, customer retention or economic impact. 1

The more important signal is the programme design. AI appears beside digital identity, payments, connectivity, cybersecurity, digital governance, public services and finance. A service becomes commercially useful only when those systems let a customer reach it, prove eligibility, exchange value, trust the outcome and complete the next action.

A population is not a market. A model is not a service. Access, trust and a completed outcome create demand.
02

Why the signal travels

The same constraint appears across developing economies

The World Development Report 2026 argues that developing countries do not need to build frontier models to benefit from AI, but importing tools is not enough. Connectivity, compute, local context, skills, institutions and a functioning business environment determine whether organisations can adopt and adapt them. 2

A separate World Bank programme assessed cloud and data-infrastructure conditions across 15 priority countries and concluded that supply, demand and rules must be developed together. Data centres and connectivity without customers create stranded capacity; demand without reliable infrastructure fails to scale; both remain fragile without data protection, interoperability, cybersecurity and workable procurement. 3

03

The denominator

Measure reachable demand, not theoretical reach

ITU's Africa-region statistics cover 44 economies. In 2024, mobile broadband covered 86% of the population, yet only 38% used the internet. Urban internet use reached 57%, compared with 23% in rural areas. These are regional figures from a defined period, not Nigeria's current market size, but they show why network coverage cannot stand in for a reachable customer. 5

UNDP reports that more than 500 million Africans still lack a verifiable digital identity. For a service that depends on eligibility, account opening, payments or trusted records, that is not a social-policy footnote. It changes conversion, fraud, support cost and who can participate. 4

From headline market to reachable demand
POPULATION

Theoretical reach

People who could conceivably benefit from the service.

Market narrative
ACCESS

Serviceable reach

People with the device, connection, identity, payment and language path.

Operating denominator
OUTCOME

Viable demand

Customers who repeatedly complete a valued outcome at workable economics.

Investment gate
04

Where value accrues

The winning product may be a bundle of complements

Value may accrue to the model provider, but it may also sit with the operator that provides connectivity, the payment network that completes the transaction, the institution that owns the trusted workflow, the local firm that adapts language and support, or the government that creates anchor demand through services and procurement.

That changes partnership economics. Do not negotiate only a model price and reseller margin. Map who supplies the customer, identity, payment, data, support, authority and last operational step. Then decide which participant earns revenue, carries cost and owns the customer relationship. A market-entry plan that leaves these transfers invisible will overstate both demand and margin.

Put complement costs beside product revenue in the same investment case. Include customer acquisition, identity verification, payment failure, local support, data preparation, training and regulatory operation. Release capital in stages as anchor demand and repeat use appear. If the economics require a partner or public institution to absorb a cost indefinitely, describe that dependence explicitly rather than reporting it as a product margin.

05

Regional comparison

Brazil shows a different starting point

OpenAI's Brazil expansion begins from substantial existing use. The company says 35% of classified messages from individual Brazilian accounts were work-related in June 2026, compared with 30% globally. That is supplier analysis of individual-account messages, not enterprise return or representative national adoption. 7

The operating response is still local: commercial staff in São Paulo, Portuguese developer events, mobile-first training with an SME institution and a public-service memorandum with the city. The same five complements apply, but the starting assets differ. In the Middle East, sovereign infrastructure and Arabic capability are prominent; in Europe, data and rights rules weigh more heavily; Australia offers mature connectivity and public assurance but concentrated cloud dependence.

06

Scale across borders

A large regional market needs systems that work together

The World Bank's 28 August brief on African integration argues that the economic value of continental frameworks depends on the systems firms use every day: customs, transport, standards, payments, energy, finance, digital platforms and data. These systems often remain fragmented across national borders. 6

For an AI service, regional scale is therefore not a translation exercise. It is an interoperability and operating-model problem. Price the cost of country-specific identity, payments, data movement, support and compliance before claiming a regional addressable market. Fund shared rails where they improve several services; sequence country entry where they do not.

07

The next 30 days

Build the market-formation account

Choose one planned AI market entry. Replace the top-down population estimate with three denominators: people or organisations in scope, those who can reach and complete the service, and those who repeat a valued outcome. Record the date, geography and definition behind each figure.

Then budget five complements: anchor demand, delivery rails, local context, operating capability and rules. Name the partner and owner for each, the evidence that it exists and the condition that releases the next tranche of capital. Stop, narrow or sequence the entry when a critical complement has no owner or its cost breaks the unit economics.

The 30-day request: one market, three honest denominators, five funded complements and one explicit stop condition.

Research record

Method and limitations

Method

This executive brief compares the current GITEX Nigeria programme with official World Bank, UNDP and ITU work on AI readiness, cloud-market development, digital public infrastructure, connectivity and regional integration, plus a current supplier expansion case in Brazil. Event counts, regional statistics, programme findings and supplier data retain their definitions and limits. The newest arXiv batch was reviewed as an underlying R&D radar and did not displace the stronger market-formation signal.

Limitations

Event participation does not establish adoption or economic impact. ITU figures describe a 44-economy region using 2024 data and cannot size Nigeria in 2026. UNDP and World Bank materials synthesize diverse countries and programmes. OpenAI's Brazil data is supplier-authored and based on classified individual-account messages. Market economics, identity, payment, language, infrastructure and regulation vary materially by country.

First published 1 September 2026 · Updated 1 September 2026 ·Research period August 2026 – September 2026 · Research current to 1 September 2026 · Version 1.0 · Suggested citation: The AI Institute, The Next AI Market Is Built Before It Is Sold (2026).

References

References and source notes

  1. 01
    GITEX Nigeria 2026 official programme

    Official programme and organiser counts for 31 August–3 September 2026; not adoption proof.

  2. 02
    World Bank, World Development Report 2026

    Current multilateral synthesis on AI in developing economies and the complements required for value.

  3. 03
    World Bank, Building Data Infrastructure for AI Readiness

    Programme results across 15 priority countries; coordinated supply, demand and rules.

  4. 04
    UNDP, Africa Accelerator for Digital Public Infrastructure

    Official launch and policy findings dated 29 June 2026.

  5. 05
    ITU, State of digital development and trends in Africa

    Official 44-economy regional report using 2024 connectivity and use observations.

  6. 06
    World Bank, Integrating Africa: From Threads to Hubs

    Regional-integration brief published 28 August 2026.

  7. 07
    OpenAI, Expanding our presence in Brazil

    Supplier announcement dated 27 August 2026; usage analysis is not enterprise ROI.

  8. 08
    ITU, ICT Development Index 2026

    Official index covering 159 economies and persistent affordability, use and quality gaps.

Download

Download the paper.

Get the print-ready PDF and receive future Institute research by email.

This web page is the accessible version of record.