The World May Spend $1 Trillion on AI This Year. Your Business Probably Does Not Need a Robot Receptionist

Global companies are expected to invest more than $1 trillion in artificial intelligence during 2026, according to Goldman Sachs Research. The estimate includes around $581 billion in US investment and captures spending by hyperscalers, other public companies, private firms and AI-exposed companies outside the United States. The widely quoted estimate of roughly $794 billion in hyperscaler capital expenditure may actually understate total global AI spending by approximately $200 billion.

One trillion dollars is a difficult number to visualise. It is enough money to make every startup founder add “AI-powered” to a pitch deck, including businesses whose main innovation is a slightly faster way to deliver plantain chips.

The investment is not going only into chatbots. It is financing data centres, chips, electricity generation, cloud infrastructure, cooling systems, cybersecurity, models, software and the specialised equipment required to make AI work reliably. Morgan Stanley estimates that data-centre construction linked to AI could require approximately $2.9 trillion through 2028, with more than 80% of that investment still to come.

This makes AI less like a fashionable application and more like an industrial buildout. Somebody must manufacture chips. Somebody must build data centres. Somebody must supply power. Somebody must cool the equipment. Somebody must protect the systems. And, eventually, somebody must produce enough revenue to justify all the spending.

That final part is becoming important. Investors have spent several years rewarding companies for talking about AI. The next stage will reward companies that can show measurable gains: lower costs, faster production, higher sales, better risk management or improved customer service.

For African SMEs, the lesson is not to compete with global technology companies in building foundation models. A company in Yaoundé does not need to construct the next global language model between resolving a tax issue and finding fuel for the generator.

The opportunity is application.

A logistics company can use AI to plan routes and anticipate delivery delays. A retailer can analyse sales patterns and improve inventory decisions. A school can personalise revision materials. A clinic can improve appointment management and administrative records. A food processor can forecast demand and reduce wastage. A media company can accelerate transcription, clipping and content distribution.

Yango’s recent claims offer a practical local example. The company reported that AI-assisted route optimisation saved Cameroonian users a combined 136,026 hours in 2025, including an average of 81 minutes per user annually in Yaoundé and 63 minutes in Douala. The value proposition is not “we have artificial intelligence.” It is “people recover time.”

That distinction is everything.

Customers do not buy AI. Customers buy cheaper services, shorter waiting times, fewer errors, better recommendations and more reliable delivery. The acronym is merely inside the machine.

Investors evaluating AI-related businesses should ask simple questions:

1. What expensive problem does the technology solve?
2. Is the underlying data reliable?
3. How much does the system cost to operate?
4. Does it improve revenue or margins?
5. Can customers tell the difference?
6. What happens when the model is wrong?
7. Does the company have permission to use the data?
8. Is there a cheaper non-AI solution?

AI investment also creates second-order opportunities for Africa. Data centres need electricity, land, fibre connections, security and cooling. Local language systems need African datasets. Companies need professionals who understand both technology and sector-specific problems. This could generate markets for power developers, cybersecurity companies, training institutions and data-service businesses.

The risk is that African companies become permanent consumers of foreign AI rather than owners of useful data, applications and infrastructure. If every business sends its customer information into systems it does not understand, the continent may gain productivity while losing strategic control.

The Summith takeaway: The trillion-dollar AI story is not an instruction to rename every business “something.ai.” It is a reminder that capital follows measurable productivity. Start with the problem, calculate the value, protect the data—and only then invite the robot.