Ethiopia markets wager on grid uptime ahead of COP32

Ethiopian Electric Power has 18 grid projects on the clock, spanning mobile substations and transformer capacity upgrades, with the finish line set at COP32 in Addis Ababa in November 2027. Look at what that deadline actually is. A summit date, not a demand forecast.

Ethiopia markets do not need another ribbon cutting. They need to know what happens on the Tuesday after the delegates fly home.

The deadline is political, the load is not

Binding 18 projects to one calendar event sets up the obvious risk. Capacity gets sequenced to a conference, not to where factories, cold chains and server halls are genuinely drawing power. This is how a grid can look complete on a map while one industrial park still runs diesel backup through peak hours.

Watch which projects land first. Transformers and substations tend to arrive where the cameras are. That is not a conspiracy, it is just how capital projects behave when the reporting line runs through a summit rather than through a regulator's reliability targets.

The second-order effect is subtler. Mobile substations are the flexible piece of the programme, which sounds like good engineering until you ask who owns them and who can move them. A trailer-mounted unit is capacity on loan. If EEP can redeploy it, it can redeploy it away from your site.

Rented capacity is a churn machine

Here is the lens I keep coming back to. Mobile substations are the grid equivalent of a subscription plan: fast to deploy, low capital commitment, renewable at the provider's discretion. Fine for the utility's balance sheet. Rough for anyone building a five-year case on top of it.

Uptime is the product. If uptime is rented, reliability is rented. Industrial buyers price that into their own supply contracts, and the ones who cannot wait buy generators instead. That is the quiet leak in every state utility's revenue base: the strongest customers fund their own way off the grid, then the fixed costs get spread across the customers who stayed.

Ask any SME what a dedicated transformer costs them, upfront, before a single unit of output leaves the gate. The answer tells you whether reliability will be shared or simply purchased by those who can afford it. Subscription economics only work when the subscriber can cancel. Here, the industrial user cannot switch provider and cannot port capacity elsewhere. Switching cost is total, which is fine right up until the service fails. Then churn stops being a metric and becomes a shutdown.

What the summit has to prove

The real test is credibility. COP32 will be judged by people who write about energy for a living, in a city whose grid is the host's own exhibit A. One blackout during the conference does more damage to the investment case than any number of commissioned substations.

The risk is not that these 18 projects fail. The risk is that they succeed at the wrong thing: enough capacity to keep Addis lit for a fortnight, and no durable answer on how the cost is recovered. Whether that recovery lands on tariffs, on the budget, or on the industrial customer who paid for the connection, is the question nobody in the announcement has answered. Reliability that ends when the visitors leave is not infrastructure. It is set dressing.

My expectation: the mobile units stay connected long after November 2027, the tariff question stays open, and the customers with the most to lose keep buying generators in the meantime.

Before underwriting any Ethiopian industrial load, ask two things. Where does this power physically come from, and who has the authority to take it away? Eighteen projects is a serious programme. Running it to a conference date shifts the risk onto the buyer. Price it that way.