Summary

  • Helios Towers reported H1 2026 tenancies of 34,455, up 13% year on year, a tenancy ratio of 2.26x against 2.11x, revenue of US$466.3m (up 11%) and adjusted EBITDA of US$257.0m (up 14%, a 55% margin).
  • The same period produced a 16% fall in cash generated from operations to US$182.2m and reported free cash flow of US$2.4m, against recurring free cash flow of US$105.8m that the company presents as its headline cash measure.
  • Net leverage of 3.4x is 0.4x better than a year earlier but unchanged against 31 December 2025, because net debt rose to US$1,773.4m while the ratio's denominator, annualised adjusted EBITDA, rose with it.
  • The US$5,895.1m contracted-revenue backlog is undiscounted and computed on five explicit assumptions, including no fee-rate escalation and no automatic renewal; it is a disclosure of contractual structure, not evidence of cash collected.
  • Power cost US$100.9m, or 21.7% of revenue, is the single largest operating input, and the company's protection against it — contract escalators it describes as effective — is asserted rather than quantified.
  • The 2027 convertible bond is pre-funded by an undrawn US$250m term loan, but the refinancing step is funded, not resolved: whether it is retired with cash, refinanced or converted depends on the share price and on the free cash flow that has not yet appeared.

The state difference is conversion, not demand

Helios Towers came into 2026 with a simple operating proposition: an existing tower costs almost the same to run with one tenant as with two, so each additional lease on a built site drops through to EBITDA at a marginal cost close to zero. That proposition was visibly working in the half-year to 30 June 2026, announced on 30 July 2026.

Sites ended the period at 15,270, up from 14,515 a year earlier. Tenancies reached 34,455 from 30,617, a 13% increase; the tenancy ratio, the average number of tenants per tower, moved to 2.26x from 2.11x and from 2.17x at the 2025 year end. Total colocations rose to 19,185, comprising 13,999 standard and 5,186 amendment colocations. Revenue reached US$466.3m, up 11% from US$418.3m, "largely driven by tenancy growth", and adjusted EBITDA rose 14% to US$257.0m at a 55% margin, up from 54%. Operating profit was US$162.9m, up 22%, and return on invested capital improved to 14.4% from 13.6%. Those are the figures the H1 2026 results announcement leads with, and the regulatory filing copy on the FCA National Storage Mechanism carries the same tables.

The cash statement tells a different story about the same six months. Cash generated from operations fell 16% to US$182.2m from US$216.0m, which the company attributes predominantly to working-capital movements. Recurring free cash flow — portfolio free cash flow less net interest paid and the net change in working capital — rose 52% to US$105.8m, or 10.2 cents per share from 6.6 cents. Reported free cash flow, which additionally charges discretionary capital additions and cash paid for exceptional and EBITDA-adjusting items, was US$2.4m against US$29.9m a year earlier.

The company is not misrepresenting either measure; it defines both, and the investor presentation accompanying the results sets out the reconciliation explicitly. Adjusted EBITDA of US$257.0m less maintenance and corporate capex of US$12.8m, less lease liability payments of US$26.3m and less tax paid of US$27.0m produces portfolio free cash flow of US$190.9m; net interest and working capital take that to US$105.8m of recurring free cash flow; discretionary capital additions of US$101.7m take it to US$2.4m.

That ladder is the whole article in one line. Each rung is a real claim on the business — leases, tax, interest, growth capital — and the last rung is where the shareholder's cash would sit if the company stopped building. The gap between US$105.8m and US$2.4m is not an accounting artefact; it is the price of the tenancy growth that produced the 14% EBITDA increase. The unresolved question is whether that growth was bought with capital that the cash flow cannot yet self-fund.

Where a colocation becomes cash

The mechanism behind a tower company's economics can be traced step by step, and Helios Towers' disclosures allow most of it to be priced.

A new tenant on an existing tower adds contracted revenue with no new site. The incremental cost is mostly electricity for the tenant's equipment, any additional ground-lease or amendment payment, and maintenance. Helios Towers' cost of sales in H1 2026 was US$212.5m, or 45.6% of revenue, down from 50.4% a year earlier — the operating leverage is visible. Within that, power cost US$100.9m, or 21.7% of revenue, is the dominant input; non-power cost of sales was US$50.0m, or 10.7%; site depreciation was US$61.6m, or 13.2%.

Power is therefore the swing factor in margin quality. The company states that its customer contracts embed annual CPI escalators and quarterly or annual power-price escalators, and describes them as providing effective protection from inflation and power-price movements on power and non-power costs. Read precisely, that is a contractual claim about escalation clauses, and the company does not publish a quantified reconciliation of escalator revenue against actual opex changes. In markets where diesel and grid tariffs move faster than contract resets, the protection is a timing mechanism as much as a hedge.

The company's own commentary describes a mix of roughly 17 hours of grid supply a day with the remainder covered by solar-hybrid and fuel systems, backed by several months of fuel — a resilience claim, and also a statement that fuel exposure has not been eliminated.

Below EBITDA sit ground leases. Lease liabilities at 30 June 2026 were US$235.3m, of which land was US$29.6m short-term and US$195.8m long-term. The maturity disclosure shows total undiscounted lease contractual payments of US$759.3m: US$43.5m within a year, US$146.3m in two to five years, US$154.8m in six to ten years and US$414.7m beyond ten years. Critically, the company states that these undiscounted cash flows exclude CPI and other index escalations. The largest single cost after power is therefore disclosed in a form that is knowingly incomplete: the base rent is visible, the escalation on top of it is not quantified.

Any model of long-run free cash flow has to supply that number, and the company has not.

Then come interest and tax. Tax paid was US$27.0m in the half. Interest is now governed by a cost of debt the company puts at 6.7% after April 2026, when it issued US$500m of 6.750% senior notes due 2031 to refinance its 2028 term loan, cutting its cost of debt by about 40 basis points and extending weighted-average maturity by a year to roughly four years. That is a genuine improvement in the terms on which the business rents capital — and it is still a 6.7% hurdle on US$1,980.0m of gross debt, which is roughly US$130m of annual cash interest before any change in the mix.

Finally, growth capital. Discretionary capital additions were US$101.7m in the half, funding 2,511 tenancy additions year to date, including 524 sites. The company's own unit economics for its guidance uplift imply a US$10m annualised increase in adjusted EBITDA and recurring free cash flow for an additional 500 tenancies, or about a 30% return on invested capital on that incremental growth. If that return is real, the capital is well spent and the US$2.4m of reported free cash flow is the arithmetic consequence of choosing growth over distribution.

If the return is thinner, the same arithmetic reveals a business that is consuming the cash it generates.

Contracted revenue is a disclosure, not a collection

The number that carries the most rhetorical weight in the release is the contracted-revenue backlog: US$5,895.1m at 30 June 2026, undiscounted and measured over the life of the contracts, with an average remaining initial life of 6.5 years, down from 6.8 years. Of that, US$5,776.9m, or 98.0%, is attributed to large multinational mobile network operators and US$118.2m, or 2.0%, to other counterparties; about 70% is from customers the company identifies as investment grade, naming Airtel Africa, Vodacom, Orange and Omantel.

The profiled drawdown is US$453.5m in the six months to 31 December 2026, US$828.1m in 2027, US$803.9m in 2028, US$750.5m in 2029 and US$707.3m in 2030.

Those figures are useful precisely because the company states five assumptions behind them: no escalation in fee rates, no increases in sites or tenancies other than committed ones, no utilisation of customer cancellation allowances, no early termination of master services agreements and no automatic renewal. In other words, the backlog is the minimum nominal contract value assuming customers behave exactly as contracted and nothing escalates. It is not a forecast, and it is not a cash-flow statement.

The distinction matters because 2027's US$828.1m is roughly 89% of H1 2026 revenue annualised, and the entire 2026 guidance range of US$520m–US$535m of adjusted EBITDA sits within the first two years of that profile — the backlog is a measure of visibility, not of surplus.

The concentration point is equally double-edged. That 99.6% of H1 2026 revenues came from multinational mobile network operators is a quality signal: these are large, mostly investment-grade counterparties with long contracts. It is also a single-industry, small-number customer base whose demand depends on mobile data competition in markets where consolidation is a live policy question. The company did not disclose individual customer revenue shares in the H1 2026 materials, so the true concentration of the portfolio is not publicly quantified — a material disclosure gap for anyone assessing tenant-consolidation risk.

Currency is the other structural constraint. The company reports that 69% of revenue and 71% of adjusted EBITDA were denominated in hard currency — US dollars, the CFA francs pegged to the euro, or the Omani rial pegged to the dollar. That is a strong figure by African infrastructure standards and it leaves roughly three-tenths of the business exposed to local-currency movements against dollar-linked debt. With gross debt at US$1,980.0m and revenue of US$466.3m in the half, a currency-driven shortfall in the non-hard-currency third lands directly on the leverage ratio rather than being absorbed by a cushion of excess cash flow.

The leverage ratio depends on which comparator you choose

Net leverage of 3.4x is presented as improving, and against the correct same-period comparator it is: 3.8x at 30 June 2025, a fall of 0.4x. The company is explicit that the measure is net debt divided by annualised adjusted EBITDA, calculated per the Senior Notes definition.

Three facts sit alongside that improvement.

First, against the most recent balance-sheet date the ratio has not improved at all. Net leverage was 3.4x at 31 December 2025 and 3.4x at 30 June 2026. Over that half-year, net debt rose to US$1,773.4m from US$1,723.3m, gross debt rose to US$1,980.0m from US$1,940.6m and cash fell to US$206.6m from US$217.3m. The ratio held only because annualised adjusted EBITDA rose to US$519.1m from US$502.1m. The company can fairly point to EBITDA growth as the cause of stability, but a reader should notice that the numerator moved the wrong way while the ratio stood still.

Second, the denominator is an annualisation. Annualised adjusted EBITDA doubles the half-year figure, which assumes the second half repeats the first in seasonality and in mix. It is the industry convention and it is also an assumption embedded in the number that determines how much financial risk the balance sheet carries.

Third, the directors' going-concern statement is qualitative. They state that the group should remain adequately liquid and should operate within the covenant levels of its current debt facilities, having considered variable energy prices, broader inflation and the geopolitical environment. No numeric covenant ratio or headroom quantum is disclosed. For a company with US$1,980.0m of gross debt, the absence of a quantified headroom figure means the covenant position must be inferred from the leverage ratio rather than observed.

Refinancing: funded is not the same as resolved

The maturity profile has been actively managed, and the sequence is worth setting out in order. In April 2026 the group issued US$500m of 6.750% senior notes due 2031, refinancing its 2028 term loan, cutting the cost of debt by about 40 basis points to 6.7% and extending weighted-average maturity to roughly four years. In May 2026 it raised a US$250m three-year term loan to manage its 2027 convertible bond and for general corporate purposes; that facility remained undrawn at 30 June 2026.

The 2027 convertible is a US$300m 2.875% instrument with US$180m outstanding and, as stated in the filing, an exchange price of US$2.9312; it is accounted for as a compound instrument with US$148m treated as a liability and US$32m as an equity component before transaction costs. The group reported more than US$500m of cash and available debt facilities at the period end, and Moody's upgraded it from B1 to Ba3 in February 2026.

What this achieves is important and limited. The 2027 maturity no longer depends on the market being open in 2027: the money is raised and undrawn. But an undrawn facility is a commitment to pay, not a discharge of the obligation. Whether the convertible is retired with cash, refinanced at a higher coupon, or converted into equity depends on the share price relative to the exchange price and on the free cash flow available to redeem it. The company has removed timing risk from the 2027 step; it has not removed economic risk from it.

And the wider maturity wall has not disappeared — the April 2026 notes push the largest scheduled repayment to 2031, at a coupon of 6.750% that is now the group's marginal cost of long-dated debt.

Returning capital before leverage has fallen

Capital allocation is running ahead of deleveraging. In H1 2026 the company completed US$27m of share buybacks and a further US$7m to 24 July 2026, taking cumulative buybacks to US$58m since the programme began in November 2025. It declared an inaugural interim dividend of 0.604p gross per ordinary share, payable 14 September 2026 to holders on the register at 7 August 2026. Full-year 2026 guidance contemplates a US$51m buyback and a US$25m dividend — US$76m of distributions — alongside discretionary capex of US$215m–US$245m.

The question this raises is not whether the distributions are affordable, because they plainly are at the stated cash levels, but whether they are the best use of the last dollar of fungible cash at 3.4x leverage. The company's IMPACT 2030 framework answers prospectively: more than US$1.3bn of cumulative recurring free cash flow targeted across FY 2026–30, net leverage of 2.5–3.5x and more than US$400m of investor distributions, with accretive organic investment taking more than US$500m. Read as a whole, the plan intends to grow, pay down and distribute simultaneously, funded by recurring free cash flow that rose 52% in the half.

That is coherent if recurring free cash flow is the right measure of capacity. It is less obviously coherent if the discretionary capital spending that turns recurring free cash flow into reported free cash flow keeps rising as guidance implies: full-year discretionary capex guidance was lifted to US$215m–US$245m from US$180m–US$210m at the same time as tenancy-addition guidance was raised to 3,500–4,000 from 3,000–3,500, with the company noting that non-discretionary capex is unchanged at about US$50m. Growth is the priority, and the distributions are calibrated to co-exist with it rather than to accelerate the debt reduction.

What would settle the question

The deleveraging thesis is falsifiable with the company's own disclosure. It would be supported if the operating-cash-flow shortfall reverses in the second half and full-year reported free cash flow converges toward the US$220m–US$235m of recurring free cash flow now guided; if net leverage moves below 3.4x against the 31 December 2025 comparator, not only against older year-on-year comparators; if discretionary capex falls as a share of tenancy additions; and if the company publishes a quantified reconciliation showing that CPI and power escalators offset the escalations excluded from its lease-payment disclosure.

It would be undermined if working capital absorbs cash for a second consecutive half; if net debt continues to rise while the leverage ratio is held flat by annualisation; if a major tenant consolidates or renegotiates, given that no individual customer share is disclosed; if local-currency depreciation in the roughly 31% of revenue that is not hard currency drives reported leverage higher; or if the 2027 convertible is addressed at a materially higher cost than the 2.875% coupon it carries.

The next hard data points are the full-year results, the level and composition of discretionary capex within them, and any disclosure of numeric covenant headroom. Until then, the honest summary of Helios Towers' H1 2026 is that the operating engine is working and the cash engine is still catching up, and the difference between those two statements is the risk an investor is being paid to carry.

Sources and evidence boundary

This article is built on Helios Towers' own filed and published material for the six months ended 30 June 2026, announced 30 July 2026, together with the prior-year comparative release.

Where a figure is the company's own characterisation rather than a reported outcome — the "effective" contract protection against power and inflation, the implied 30% return on incremental investment, the IMPACT 2030 targets — it is attributed as such. No individual customer revenue share, no numeric covenant headroom and no quantified reconciliation of escalators against opex were disclosed in the material reviewed, and those gaps are stated rather than filled. Directory entry: Helios Towers.