GO internet — Adding fibre and 5G capacity

Historically, subscriber growth has been aligned to the pace at which GO installs base stations (see Exhibits 9 and 10). While it stepped up the pace at which it installed its 100Mbps LTE base stations during FY17, churn from the 12MBps WiMAX network is affecting overall growth, capacity remains stretched and net subscriber additions, at 3.7k in FY17 (compared to 6.4k in FY16), were subdued.

The strategic agreement with Linkem will see the two groups combine frequencies in the Marche and Emilia Romagna regions. This will enable GO to significantly increase both capacity and service levels. In our forecast, we assume the new network is installed over a three-year period starting in 2019, enabling GO to retain and grow its wireless customer base from 41k in 2017 to 53k by 2020. For the new fibre services, we extrapolate GO’s experience from the trial in Perugia to the other cities, which assumes a market share of 1% (7,000 subscribers) by 2019. With the mix of subscribers shifting to the higher-speed services, we also expect to see improvements in ARPU. Together, this should drive an accelerating revenue growth profile in FY18e (14%) and FY19e (22%), translating to a 20% three-year CAGR in EBITDA.

While the capital expenditure requirements for its fibre plans are fairly low, we forecast that the installation of the new 5G network will require capital expenditure to increase from €3.6m in FY18e to €5.5m in FY19 and €6.1m in 2020. Net debt was €4.5m in December 2017. We expect management to refinance €3m of this debt that matures over the coming three years. However, the €4m from Linkem would enable the network build to be completed on a lower level of gearing, which we forecast would peak in 2020 at a comfortable 1.2x EBITDA.

With growth in subscribers slow to pick up, GO’s shares marked time during 2017. Reverse DCF suggests that the current price discounts a continued slowdown in the rate at which it can add subscribers to the wireless network, with nothing in the price for fibre. However, there are a number of factors that could provide upside to this and investors should consider the following:

Our forecasts assume a stable pricing environment and steady progression in subscribers. However, the market is highly competitive and GO needs to react to market trends. The success of the fibre service may not be consistent with the outcome in Perugia and the timing, cost and uptake of the deployment of the 5G overlay network may vary to forecasts.

GO internet provides fixed wireless internet services using 4G technology in the Marche and Emilia-Romagna regions of Italy, where it owns the licences for the exclusive right of use for 42MHz in the 3.5GHz frequency band. With its wireless network, it is targeting the approximately one million homes in its licensed regions that have opted to relinquish their fixed-line connections and, consequently, have no fixed broadband service, along with homes in the region that have no DSL access. Its original WiMAX kit offers speeds of up to 12Mpbs; however, this network is at capacity and since 2015 it has been installing LTE kit that offers speeds of up to 100Mbps, which has been the main driver of subscriber growth over the last two years.

In March 2018, it announced an agreement with Linkem where they will combine their frequencies within Marche and Emilia-Romagna giving access to 84MHz. This will enable it to install an overlay network to facilitate the migration of its WiMAX subscribers to 5G as well as providing additional headroom for new subscriber growth.

In addition, following a successful trial in the city of Perugia (Umbria), GO recently announced that as part of a broad agreement signed with Enel Open Fibre (EOF), it will also start to offer 1Gbps fibre-to-the-home (FTTH) broadband services to a further 22 cities within Marche and Emilia-Romagna.

GO’s subscriber base has expanded steadily since its IPO in 2014. In total, GO has 42.6k subscribers (February 2018), of which we estimate 74% are WiMAX (12Mbps), 25% LTE (100Mbps), and approximately 1% accounted for by the fibre pilot in Perugia.

WiMAX and LTE are wireless technology standards that come under the umbrella of 4G services. Typically associated with mobile phone services, the technology can also be used for “last mile” connectivity (the most difficult and expensive to build for wire line services) to provide a fixed broadband service.

GO owns the licences for the exclusive right of use for 42MHz in the 3.5GHz frequency band in the Marche and Emilia-Romagna regions. The frequency-sharing arrangement with Linkem means that it will, however, have access to 84MHz. GO acquired its first licence in 2008 and its second in 2012; both expire in 2023. In Q118 the Ministry of Economic Development accepted GO’s application for an extension to its licences for an additional six years for €1.6m (the timing of the payment is not yet decided).

Developing a fixed wireless broadband network requires the installation of a base station (which cost €5–8k) on a networking tower, which can then connect to a customer’s home via customer premises equipment (CPE) or modem (as illustrated in Exhibit 3). Each base station can support 50 to 60 customers, although the newer 5G compatible base station technologies that GO is proposing to install have the potential to support significantly more.

GO started rolling out its network in 2011. The original WiMAX kit installed can support speeds of up to 12Mbps. This part of the network is now at capacity, with speeds that are no longer competitive. From the summer of 2015, GO moved to the installation of 100Mbps LTE base stations and is currently installing base stations that support speeds to 300Mbps.

Of the c 925 base stations installed at the end of December 2017, 412 are WiMAX only, 383 are 100Mbps LTE, of which just under half are dual mode and could support 300Mbps (subject to software upgrades) and a further 130 are dual mode WiMAX-LTE (100Mbps-capable following software upgrades).

Linkem deal: by sharing frequencies with Linkem, GO will be able to install more powerful 5G technology (speeds of up to 1Gbps), which should enable it to both migrate its WiMAX subscriber base to a better service level (reducing churn) as well as almost doubling its capacity in the region, given 5G base stations support a higher attachment rate of subscribers. As part of the agreement, Linkem will subscribe to €4m of new equity, which at €1.4 per share represents an additional 2.9m new shares – enlarging the share capital base by 21.2%. The recapitalisation is expected to take place by the end of June 2018.

During 2015, state-controlled power utility Enel (ENEL.SW) started installing a new generation of smart electricity meters in homes across Italy and in 2016 announced that it would lay fibre in parallel. The regulator has stipulated that the subsidiary managing the fibre network (EOF) may only offer the network on a wholesale basis and EOF has partnered with various third-party service providers, including GO internet.

As demonstrated in Exhibit 4, in order to offer fibre services, GO needs to install an optical line terminal (OLT) costing c €5k into the local exchange. Once a subscriber is contracted, it will also need to install a vertical (approximately €100 per apartment building) connecting the local modem to the subscriber’s CPE (€50). While GO’s capital expenditure requirements are much lower than for its wireless network, once a subscriber is contracted GO pays a monthly rental for the last mile connection to EOF of approximately €12 per month (with a lower rate in the first 12 months) and consequently gross margins are lower. We understand that EOF plans to install fibre to approximately 900k homes over the next 12 months in the cities targeted by GO.

In Italy there is no cable network, and so the broadband market is largely ADSL based. In addition, three 4G licences for 3.5GHz spectrum have been awarded in each region in Italy: one regional licence holder, and two national. The national licences are owned by Linkem and Tiscali (which merged with licence holder Aria in 2016), and the regional ones by a range of companies, including Aolus, Air and GO internet (included in ‘other’ in Exhibit 6).

Total broadband penetration in Italy at 63% (+3pp y-o-y), continues to lag that of the rest of Europe. However, the government’s “Italian ultra-broadband and digital growth 2014-2020 strategy” approved in March 2015 seeks to address this gap. The plan targets that 85% of homes will be able to access connection speeds of 100Mbps or over by 2020 with the remainder guaranteed 30Mbps+. The government has actively pursued this target, shaking up the Italian Telecoms market in the process. Over the last 12 months, the share of the market accounted for by ultra-broadband access has nearly doubled to 20.8% (Exhibit 5).

New entrant in fibre: the most significant change over the last 18 months has been the award to a new entrant, EOF, of a multi-billion euro contract for the construction of a near nationwide fibre network (covering 271 cities). Although still too early to determine the exact consequence this will have on the market, EOF’s entry appears to have served as a catalyst to the incumbent providers, which have become more aggressive in terms of pricing and next-generation network roll-out plans. As EOF is restricted to the wholesale operation of a fibre network, it is stimulating the virtual network operator (VNO) market for fibre services.

More aggressive next-generation network (NGN) roll-out plan from Telecom Italia: TI, the market leader with a 45% broadband market share, announced in early 2017 that it had accelerated its ultra-broadband roll-out plan across Italy. It is targeting 95% NGN coverage by 2019 (from 65% in 2016) and is looking to increase its NGN customers fourfold by 2019. As part of this plan it aims to be first to market with FTTH coverage in 50 cities. It has also noticeably stepped up its content strategy and service bundling as another way to mitigate the market share erosion it has experienced over the last decade.

Increased regulatory scrutiny: during 2017, AGCOM, the Italian communications regulator, opened two enquiries into TI’s broadband roll-out. It is also reviewing the terms of the regulated TI fixed-line networks, to ensure equivalence between the terms offered to its own retail arm and third party providers. This could result in a revision to the terms on which other network operators, such as GO, pay for backhaul services or, at its extreme, in the legal separation of its fixed wholesale network (similar to the situation with Openreach in the UK).

Merger of Wind and H3G: another change has been the merger between VimpelCom’s Wind Telecomunicazioni (15% broadband share) and CK Hutchinson’s 3 Italia (H3G) to create the dominant mobile group in Italy (Wind Tre), which could result in the entry of Iliad into the mobile market, a condition for the merger’s approval.

3.5GHz spectrum to be used for 5G mobile services: the last year has seen some significant developments in determining the frequencies that will be freed up to support the gigabit technologies required in a 5G mobile environment. Regulators across Europe (Ofcom, ITU) and in Italy have identified three key bands: 700MHz, 24.5–27.5GHz and, 3.4–3.8GHz. Critically for GO, in Italy it is the 3.4–3.8GHz band that will be allocated first. While commercial 5G mobile services are still some way off, TIM is planning to launch its first test city (Turin) in 2018 and Intel, Qualcomm and Huawei have all started producing chipsets that support 3.5GHz frequency for tablet and smartphone reception, and a number of high-end phones already incorporating the technology (for instance the iPhone 8+, Samsung Galaxy S8, OnePlus 5T). According to press reports, authorities say they hope to raise €2.5bn from the sale of 5G-capable spectrum. The tender process for the spectrum for five cities in Italy is scheduled for later this year. This could open up a number of new possibilities for GO longer term, from wholesaling its spectrum to mobile operators to potentially even adding its own mobile offering to its fixed wireless broadband services. It also makes GO’s ownership of spectrum in this frequency band more valuable.

With only 63% broadband penetration in Italy, there remains headroom for growth for all operators in GO’s regions. However, unlike ADSL rivals, GO’s wireless service does not require customers to also take a fixed telephone line. Consequently, GO’s natural market to date has been the increasing number of ‘mobile-only’ homes in Italy that have opted to relinquish the Telecom Italia fixed-line phone connection, which is increasingly seen as redundant as mobile services have become more widespread (Exhibit 7).

In order to stem the declines in fixed-line subscribers, TI has reduced prices and over the last years, ‘cord cutting’ appears to have stabilised. Nevertheless, together with the c 100k homes that have no ADSL access (so called ‘digital divide homes’), there are approximately one million homes in GO’s licensed regions that for one reason or another have no fixed line connection.

GO’s brand is known for its competitive pricing, convenience and speed:

Competitively priced: without the need for a fixed-line rental, coupled with GO’s light central cost base (c 50 employees), it is able to offer competitive prices. Even taking account of special offers, GO’s services remain amongst the cheapest on the market.

High speed: the basic service runs at a maximum speed of 12Mbps with an average speed comparable with ADSL services in Italy. The 100Mbps service gives an average speed of 30Mbps, making it faster than many ADSL services, and GO will soon start to market its fibre services with speeds of 1Gbps (ie 1,000Mbps), enabling it to compete across the spectrum of service levels.

Convenience: the small and lightweight indoor CPEs, which account for the majority (c 85%) of subscribers, provide a no-hassle plug & play installation. This compares to ADSL customers, who often need to wait a few days for activation and may also require installation by an engineer.

Mobility: the plug & play modem can be used anywhere in the coverage region; for instance, customers can take it to second homes for the weekend.

The introduction of more powerful base stations is a key pillar in management’s strategy to increase coverage in its two regions and to remain competitive. Management targets the installation of c 10-15 base stations a month. Each base station can support approximately 50-60 subscribers, and the attachment rate has been fairly consistent (Exhibits 9 and 10). The lack of clarity over funding during 2016 (refer to our report Rights issue for accelerated network roll-out for more detail) meant that GO had to slow the pace at which it rolled out its network to conserve liquidity. The October 2016 rights issue, which raised an additional €4m, enabled the pace of network expansion to subsequently pick up. However, with the network nearly at capacity, and increasing churn from the WiMAX subscriber base, it has resulted in a knock-on effect on the rate at which it has signed subscribers.

Pricing is being used to encourage new subscribers to opt for the faster service (where the group has more capacity). Management periodically splits out subscriber growth by technology and this policy is clearly working, with all the new subscribers in recent months coming from its higher bandwidth services, and the majority of cancellations from the WiMAX services. Provided churn for the LTE services remains low, this should also result in an increasing ARPU as more customers enter the second year of their contracts on higher prices. GO is pre-emptively contacting subscribers reaching the end of their contract period to encourage a high rate of retention.

The installation of a 5G network overlaying its current WiMAX network would both improve the capacity of its network (these base stations should be able to carry c 100 subscribers each), as well as improving the service level considerably, offering ‘fibre-like’ speeds.

In July 2016 GO announced that it had partnered with EOF in Perugia (Umbria) to market its ultra-wideband, FTTH network. Since launching in June 2017, the group has made approximately 50 fibre subscriber additions each month and has now announced its intention to extend the offer to 23 cities.

While the initial results in Perugia are respectable (c 0.7% market share within six months) GO’s brand is less well known in Umbria. The additional cities being targeted will be in GO’s existing wireless regions, where its brand is well established. This will enable it to keep the cost of marketing to a minimum, and leverage existing retail relationships.

The 23 cities represent approximately 900k homes and GO’s marketing will track the fibre roll-out plan of EOF, which we understand targets full coverage by the end of 2018 in these areas.

We have updated our model to reflect the following significant changes:

The above changes result in a significant reduction to our previous FY18 EBITDA forecasts. However, given the accounting changes, the new fibre strategy, and the overlay network, this comparison to our previous forecasts is no longer meaningful.

We summarise our KPI assumptions in Exhibit 11 below. The core assumptions underlying our forecasts for the evolution of the wireless and wireline network include:

Wireless: we assume prices remain flat with a modest increase in ARPU coming through from 2019, when an increasing share of subscribers rotate on to the higher rate in the second year of their contracts and as fibre subscribers start to make a more meaningful contribution to the total. We assume c 75 5G base stations a year are installed from 2019, enabling the redundancy of the WiMAX network by 2022. Given the current high capacity utilisation, we assume that subscribers grow at a slower pace than the network expansion. Overall, we forecast wireless subscribers to increase from the c 41k reported at the end of 2017 to 53k by 2020.

Fibre: we base our fibre forecasts on GO’s trial results in Perugia, although assume a relatively small increase to the total marketing budget given the savings that can be made by marketing both the wireless and fibre products jointly. We assume EOF has coverage of c 0.9 million homes in these cities by 2020, and that GO can achieve an average 1.5% market share by this time. With customers contracting for four years, churn is not a factor for the forecast period and this translates to 1.8k fibre subscribers by the end of FY18 and 13.5k by the end of 2020.

In FY17 capital expenditure was €3.9m. We forecast a similar level in FY18; however, we forecast that the implementation of an overlay 5G network would result in a step-up in the level of capital expenditure to €5.5m in FY19 and €6.1m in FY20.

GO reported net debt of €4.5m at December 2017 with cash of €1.0m. Debt comprises three bank loans of approximately €3.5m (c 3.5% interest) repayable in annual instalments by 2019 and 2023 and sale and leaseback arrangements mainly with Huawei, the outstanding payments due on the licence acquisition for the Emilia-Romagna region of €0.2m (repayable in annual instalments to 2024) and other credit lines.

Approximately €3m of this debt is due over the next two years. In addition to the €4m of equity issued to Linkem, we would expect GO to refinance in order to continue to roll out its network at the pace forecast. We have assumed €4m of debt (€3m refinancing and €1m additional) finance is put in place over the next two years (at 4%).

GO’s listed peers are more mature or have a different capital structure, so we are wary of making direct comparisons, although it is worth highlighting that (assuming the Linkem deal completes) on an FY18e EV/EBITDA of 6.2x, GO trades at a 25% EV/EBITDA discount to small-cap Italian telecoms peers Retelit and Tiscali despite its higher EBITDA margins and better EBITDA growth profile (we forecast a CAGR in EBITDA of 20% from FY17 to FY20).

Given the proposed network build, we prefer a DCF valuation. Our reverse DCF suggests that the current price now discounts a continued slowdown in the rate at which it can add subscribers. However, we believe there are a number of factors that could provide upside to the current price and investors should consider the following:

Base case DCF 220c: we adopt a 10-year explicit forecast period, of which the first three years are driven from our published forecasts. We assume the overlay 5G network is completed by 2022, after which time, the number of base stations added each year stabilises at 20 with capex to sales falling from c 50% currently to 18.5% by year 10 with capacity utilisation on the wireless network stabilising at c 75% (the equivalent of 95k wireless subscribers, 21k fibre subscribers by 2028).

We factor in competitive pressures by assuming no underlying increase in the prices of the packages over the period. Overall, this translates to a 12% CAGR in revenues over the next 10 years. We assume EBITDA margins peak at 50%, a 28% tax rate. We have used a 10.5% discount rate and a 2% terminal growth rate from 2028. These assumptions return a value of €2.20 per share, 63% above the current share price. Exhibit 13 summarises the DCF based on differing WACC and terminal growth scenarios.

Strategic value: it is also important to bear in mind the potential value that spectrum ownership may have on GO’s value in light of the planned allotment of 3.5GHz technology to mobile networks as part of the government’s 5G plans. Irrespective of its eventual commercial strategy, we believe that owning these frequencies in the valuable Marche and Emilia-Romagna regions in Italy puts GO on the map of mobile operators looking to secure capacity for 5G mobile services.

Regional expansion/spectrum auctions: in early 2015, AGCOM announced its intention to release an additional 200MHz of spectrum in the 3.6GHz frequency band. In the past we believed GO was a strong contender to win some of this spectrum. In light of the proposed 5G standards, these auctions are now likely to be hotly contended, which is likely to make it harder than previously thought for GO to widen the footprint of its wireless network beyond its current regions.

As well as the sensitivities outlined above, investors should consider the following:

Additional finance: we assume €3m of refinancing and $1m new debt in or forecasts. Following the Linkem deal, and as discussed earlier, we think it unlikely that GO would opt to participate in the upcoming spectrum auctions. Should management decide to participate, which would enable it to expand its wireless network in other regions in Italy, it may seek to raise additional funds. Depending on its cost and deployment, this could have a dilutive impact on earnings.

Cost of the network: the cost of the network depends on a number of variables including the terms on which GO can secure new base stations, the cost of CPEs and the density of subscribers on each base station. Before embarking on a base station installation, GO performs market research to assess the economic viability of each site and so the risk of over-provisioning is fairly low. It is also dependent on securing frequencies.

Economy: while GO’s low price offer confers some protection, cost-conscious subscribers looking to economise in the difficult economic climate in Italy could affect demand.

Competition: the market is very competitive and more sustained discounting by the ADSL providers, or a step-up in network investment by the other 4G operators could have an impact on GO’s strategy.

Service bundling: bundling of three or more services (fixed-line telephony, broadband, mobile and TV) is now the norm in other markets. For structural reasons, Italy has been slow to follow suit (no cable). However, bundled offers have been introduced: TI has launched a pay TV over DSL service and Sky Italia has entered into a number of partnerships with broadband providers and recently announced a deal with EOF which will enable it to launch a triple play of services directly to subscribers. These offers are currently relatively expensive and GO, with its no-strings service, is targeting a different demographic. However, if bundled services start to catch on in Italy, GO may be disadvantaged.

Sites secured: Italy has some of the most restrictive rules on the installation of sites and the bureaucracy surrounding site planning applications can present obstacles to the pace of network roll-out. GO has many tower providers, but during 2015 extended its contract with one of its largest, and secured access to additional strategic towers via an agreement with EI Towers (one of Italy’s leading transmission tower companies), giving it ample capacity for the near term. The Italian government is working on a set of measures to stimulate broadband diffusion in Italy. New policies being discussed include an increase in the number of site permits made available, an increase in the speed of approval of site applications and, crucially, an increase in the electromagnetic emissions permitted per site, which would enable more base stations to be installed in each site, improving the ROI of each site installed.

Linkem deal: our forecasts assume the new network is installed over the period 2019 to 2022. There are many uncertainties regarding the timing of installation, GO’s ability to migrate its WiMAX subscribers and other variables, such as the cost of 5G base stations.