The UK government’s decision to ban Huawei entirely from its operators’ 5G networks, and to force them to remove any equipment already deployed, has thrown a harsh light on the thorny issue of rip-and-replace procedures in the mobile industry.
In the run-up to the new policy announcement, Philip Jansen, CEO of incumbent telco BT, said his firm was already set to spend £500m ($630m) to remove Huawei from the core network of its EE subsidiary, and to reduce Huawei’s presence in its access networks to 35% (the cap set, under an earlier UK policy decision, on the Chinese vendor’s share of 5G or fiber infrastructure).
To replace Huawei completely in 5G would increase that financial burden by adding up to £600m in rip-and-replace fees, risk outages and performance issues for UK users for years to come, delay full 5G availability – plus BT would need at least five years, and preferably seven, to complete the replacement.
In fact, the government has conceded on the final point, and telcos have until 2027 to eliminate Huawei equipment, which saw BT reducing its cost estimates closer to the previous budget of £500m. Meanwhile, Ericsson’s European president scoffed at the idea that it would take so many years to replace Huawei, telling Sky News his firm has carried out 100,000 site swaps in the last 12-18 months – “more than twice the number of sites across the whole of the UK”.
However, he did not comment on what it would cost to do those swaps, and who would foot the bill. Rip-and-replace is often assumed to be a major cost for the operator, but in reality, large telcos can usually get some, if not all, the cost borne by the incoming vendor as the price of entry to a new customer. Outgoing Nokia CEO Rajeev Suri has said that, while open RAN architectures such as O-RAN may appear to be an out-and-out threat to the established suppliers, they would at least save Nokia the time and money it spends on swap-outs, since open interfaces would make it simple to replace a rival’s equipment.
That is the great hope that operators and smaller vendors have for open RANs, of course. A set of fully open interfaces between different network elements in a virtualized, disaggregated network would allow physical units to be swapped in and out relatively simply, more like a carrier WiFi network, the argument goes – and meanwhile, most of the functionality would be in software, and virtual network functions (VNFs) could be spun up and down as required.
But is the timing off for this vision? Is the open RAN movement taking off just as established vendors are making their networks easier to install and replace – and with early multivendor deployments, such as Rakuten’s, looking expensive and high risk. In greenfield networks, the open approach makes perfect sense and will encourage new deployers, but most of these roll-outs – with a few exceptions like Rakuten and Dish – will be localized, often focused on enterprises or campuses, and based on small cells. There are several open interfaces which are gaining some maturity in the small cell world, particularly those from Small Cell Forum and O-RAN.
But greenfield small cell networks do not have the primary challenge that still affects legacy, macro networks – the complexity of deploying, tuning, optimizing the physical elements, which makes rip-and-replace a daunting prospect. Open interfaces do not, per se, make it simpler to roll out the antennas, radios and distributed units, which all remain specialized pieces of hardware. It may be that, as the new suppliers of open RAN equipment race to find integrator partners with these specialized skills, and to win the trust of the large operators, the big vendors will unleash the work they have been doing in recent years to simplify their roll-outs services. There will certainly be a strong incentive to do this, and so remove one of the arguments for open RAN challengers, and position themselves as the safest option for an operator which needs to replace its base stations.