What Does It Really Cost to Replace Huawei in a 5G Network?

Updated July 2026. This article is intended for network operators, infrastructure owners, procurement teams and public-sector buyers preparing an early-stage replacement budget.

The first price a buyer sees is usually the replacement equipment quote. It is also the number most likely to be misunderstood.

Removing Huawei from a live 5G network is not the same as buying a new set of radios. The job can touch antennas, baseband units, core software, network management, site power, transmission links, licences, spares, field labour and years of operating procedures. Traffic must continue to move while those pieces are changed. In some networks, 4G and 5G share hardware or spectrum, so a project described as a 5G replacement can reach older services as well.

That is why there is no credible universal answer such as “$X per site.” A limited software or core-network change, a nationwide radio access network swap and the removal of equipment from a small rural carrier are different engineering programmes. Public figures range into the billions, but those figures describe particular policies and network footprints, not a global price list.

The Short Answer

For an operator-level business case, the Huawei replacement cost can be materially higher than the price of the new hardware and software. A first budget must also account for design, installation, integration, optimisation, site remediation, dual operation, training, spares, decommissioning and schedule risk.

For national programmes, the scale can reach several billion dollars or pounds:

Market example	Published figure or timetable	What the number actually means
United States	The statutory expenditure limit for the FCC reimbursement programme was raised from $1.9 billion to $4.98 billion.	This covers eligible removal, replacement and disposal work for participating smaller providers using covered Huawei or ZTE equipment. It is not a US-wide 5G RAN price.
United Kingdom	The government estimated cumulative policy effects of up to £2 billion and a two-to-three-year delay to 5G rollout when announcing its 2020 decision; legal notices later retained the end-of-2027 removal deadline.	The figure combined several policy changes and rollout effects. It should not be read as the invoice for Huawei equipment alone.
Germany	Critical Huawei and ZTE components are to leave 5G core networks by the end of 2026, with critical management functions in access and transport networks replaced by the end of 2029.	The published timetable shows the value of phasing different network layers. The government source does not provide one aggregate replacement price.
European Union	Implementation differs by Member State under national security powers and the EU 5G Toolbox approach.	There is no single EU replacement bill because scope, installed base, legal measures and operator architectures differ.

These figures are reported in the currency and policy context of their source dates. They have not been adjusted here for inflation, exchange rates or differences in eligible scope, so they should not be compared as if they were bids for the same network.

The US numbers are particularly useful because they expose the size of costs that sit beyond a normal purchasing cycle. An FCC notice issued in January 2026 records that initial demand exceeded funding, leaving recipients with 39.5% of approved allocations before additional funding was authorised. The same notice states that Congress increased the programme limit to $4.98 billion and that further allocations were made in 2025.

In the UK, the government’s July 2020 statement on Huawei and 5G estimated a cumulative delay of two to three years and costs of up to £2 billion. The wording matters: part of that estimate related to earlier restrictions and delayed rollout, while the physical removal requirement added hundreds of millions. It is evidence of programme scale, not a per-site benchmark.

Germany illustrates a different lesson. A 2026 Bundestag summary of the government’s position separates the core-network deadline from the later deadline for critical management functions in access and transport networks. Phasing can reduce operational risk, although it may extend duplicate support and programme-management costs.

Before Pricing the Swap, Define What “Huawei” Means in This Network

An asset register may say “Huawei site” even though the site contains equipment from five or six suppliers. Another record may list a non-Huawei antenna connected to Huawei radios, a third-party microwave link, a Huawei baseband unit and a locally supplied DC power system. The label is convenient for operations, but too imprecise for a replacement budget.

The scope should be separated into four layers.

1. Core network

A core replacement may involve fewer physical locations than a RAN swap, but the migration can be more sensitive. Subscriber data, authentication, policy control, charging, lawful-intercept interfaces, roaming, network slicing and operational support functions may be involved. Cost follows software licences, data migration, testing, integration and change windows rather than tower count.

2. Radio access network

The RAN creates the large site volume. Radios, antennas, baseband or distributed units, combiners, feeders, mounts and fibre may need replacement or reconfiguration. A new vendor may not support the existing antenna arrangement, remote electrical tilt interface or baseband pooling design. The network then needs coverage and capacity optimisation after the physical work is complete.

3. Transport, synchronisation and management

Backhaul, timing, element management, fault reporting, performance counters and configuration tools are easy to overlook because they may keep working during an early lab demonstration. At scale, missing interfaces or changed alarm definitions create operating cost. The question is not merely whether the new equipment passes traffic. It is whether the operations centre can see, configure, patch and troubleshoot it without maintaining two permanent toolchains.

4. Passive infrastructure and site energy

A radio swap can change equipment weight, wind loading, cabinet space, heat rejection, peak power and average energy use. That may trigger structural checks, new brackets, rectifier expansion, battery changes, cooling work or a larger generator. On an off-grid site, even a modest continuous load increase can require additional photovoltaic modules and battery capacity.

This last layer is often treated as somebody else’s budget. It should not be. For remote, weak-grid and solar-supported sites, independent telecom site power and solar-energy engineering resources can help procurement teams frame the power audit before a radio vendor’s design becomes fixed.

The Ten Cost Lines That Belong in the Business Case

Cost line	What belongs in it	Common budgeting mistake
Asset discovery and design	Site surveys, model and revision records, configuration capture, coverage design and dependency mapping	Assuming the network inventory reflects what is actually installed
Replacement equipment	Radios, antennas, baseband or cloud infrastructure, core platforms, licences and management systems	Comparing unit prices without comparing included capacity and licence terms
Field installation	Rigging, technicians, travel, cranes, access, night work and health-and-safety controls	Applying one labour rate to city rooftops, rural towers and restricted sites
Site remediation	Steelwork, foundations, cabinet space, cooling, grounding, cabling, rectifiers, batteries and generators	Treating existing passive infrastructure as automatically reusable
Integration and optimisation	OSS/BSS interfaces, neighbour planning, feature parity, parameters, drive testing and performance tuning	Ending the budget at “equipment on air” rather than accepted network performance
Migration and dual running	Temporary licences, parallel management, traffic moves, rollback capability and extra transmission	Assuming the old network can be switched off immediately after installation
Programme governance	Security review, legal compliance, project controls, audit evidence, reporting and supplier coordination	Leaving operator staff time outside the business case
Training and spares	New test tools, field training, NOC procedures, initial spares and repair logistics	Using the old vendor’s spare ratio without checking the new repair model
Removal and disposal	De-installation, transport, storage, data sanitisation, recycling and evidence of disposal	Assuming removed equipment has resale value where rules require controlled disposal
Financing and risk	Working capital, inflation, foreign exchange, schedule contingency and service-impact exposure	Using a static equipment price for a multi-year programme

A missing access key, a frozen tower clamp or an undocumented fibre route will not appear in a vendor’s radio price. Multiply small surprises by several thousand sites and they stop being small.

A Practical Early-Stage Cost Model

When surveys are incomplete, a buyer still needs a number for funding approval. One transparent method is to express the initial hardware and software quote as 100 cost units, then add planning allowances. The following is an engineering scoping model, not a market average or supplier quotation:

• Replacement equipment and initial licences: 100 units.
• Field installation and site access: add roughly 15 to 35 units.
• Integration, migration and optimisation: add roughly 10 to 25 units.
• Programme controls, logistics and compliance: add roughly 8 to 15 units.
• Training, tools and initial spares: add roughly 3 to 8 units.
• Contingency: hold 10% to 20% of the applicable pre-contingency budget until pilot data is available.
• Site remediation, financing and service-impact risk: calculate separately because these can be negligible at one site and dominant at another.

Under those assumptions, a 100-unit equipment quote can become roughly 150 to 220 units before major site remediation or financing. That is not a universal multiplier. It is a warning against approving a national programme from a bill of materials alone.

The model should be replaced with measured data after a representative pilot. A useful pilot includes more than the easiest urban sites. It should contain at least one rooftop, one rural macro site, one constrained compound, one high-traffic cluster and, where relevant, one poor-grid or off-grid site.

Energy Consumption Can Reverse the Cheapest Bid

Capital cost gets the attention because it is visible at award. Energy cost keeps arriving after the project team has moved on.

Suppose one proposed architecture draws an average of 0.5 kW more per site than another across 1,000 sites. The annual difference is:

0.5 kW × 8,760 hours × 1,000 sites = 4.38 GWh per year

At an illustrative electricity price of €0.15 per kWh, that is €657,000 per year before cooling losses, tariff changes or demand charges. Over seven years, the undiscounted difference is about €4.6 million.

On a diesel-backed or solar-battery site, the same 0.5 kW difference has a second effect. It can increase generator run time, fuel deliveries, battery cycling and the size of the energy system required to maintain autonomy. Procurement should therefore compare watts per carried traffic load, not only nameplate power or a laboratory idle figure.

Three Replacement Strategies, Three Different Cost Profiles

Like-for-like functional replacement

The operator seeks the closest available substitute for the existing architecture. This usually limits redesign and may shorten training. It can be sensible where deadlines are tight and the network is relatively young.

The drawback is easy to miss: replacing old equipment with a similar architecture can preserve old site inefficiencies and create another upgrade programme sooner than expected.

Replace and modernise

The operator combines mandatory removal with capacity expansion, spectrum refarming, virtualised core functions or site consolidation. The project costs more than a narrow swap, but part of the spending serves an upgrade that would have happened anyway.

Financial reporting should separate three amounts: the cost caused by the removal decision, the cost of planned modernisation and the cost of bringing deferred site work forward. Without that split, replacement is blamed for every improvement attached to the project.

Introduce a multi-vendor or Open RAN architecture

A more open architecture can reduce future dependence on one supplier and widen sourcing options. It does not remove integration work. Radio, distributed and central units, cloud infrastructure, orchestration, timing and management must perform as one network. Early programmes may spend more on laboratories, system integration and field optimisation even if competition improves later purchasing.

The cheapest strategy is therefore not automatically the one with the lowest first-year capex. It is the option that meets the compliance deadline, preserves service and produces the lowest risk-adjusted lifecycle cost.

The Cost of Delay Is Real, but So Is the Cost of Rushing

A long timetable keeps duplicate support contracts, old spares, new spares and parallel management tools alive for longer. It can also postpone coverage expansion because engineers and change windows are occupied by replacement work.

A compressed timetable creates a different bill. Scarce rigging crews command higher rates. Equipment arrives before sites are ready. Warehouse and financing costs rise. Optimisation is shortened. More work moves into expensive night windows, and the chance of a customer-facing outage increases.

The UK government’s original estimate linked its policy decision with rollout delay as well as direct cost. Germany’s staged deadlines distinguish core functions from critical management systems in access and transport networks. Neither approach supplies a ready-made schedule for another country, but both show why timetable design belongs inside the financial model.

A Replacement Programme Is Not a Verdict on Product Performance

Huawei restrictions have been justified by governments through national-security, supplier-risk and supply-chain assessments. Huawei has disputed allegations and restrictions directed at the company. Those policy arguments should not be rewritten as an unsupported claim that every installed Huawei radio performs poorly or contains a proven technical defect.

For an operator, the practical distinction is important. A compliance-driven replacement may remove equipment that is still functioning and has not reached the end of its economic life. That stranded value belongs in the business case. So does the risk of continuing with equipment that has declining support, uncertain component availability or a shrinking route to authorised software updates.

The European Commission’s 2023 communication on the 5G Cybersecurity Toolbox stated that Member State decisions to restrict or exclude Huawei and ZTE were justified under its high-risk-supplier assessment. Yet implementation remains tied to national decisions and network circumstances. A buyer should price the rule that actually applies, not a headline about another market.

What Should Be Measured in the Pilot?

The pilot is where a budget stops being a spreadsheet exercise. Acceptance should cover:

• Coverage, uplink and downlink throughput at comparable load.
• Latency, call setup, dropped sessions and handover performance.
• VoLTE, emergency calling, roaming and legacy-service continuity.
• Peak and average power at the DC bus, including cooling impact.
• Battery autonomy and generator behaviour after the load change.
• Alarm mapping, performance counters and remote configuration.
• Software signing, access control, logging and update rollback.
• Actual technician hours, tower time, travel and repeat visits.
• Removed-equipment handling and the evidence required for disposal.

Record the awkward sites, not just the average. Ten difficult compounds can consume more management attention than a hundred straightforward swaps.

Eight Questions to Ask Before Approving the Budget

1. Which law, licence condition, funding rule or internal risk decision requires replacement?
2. Does the scope cover the core, RAN, management systems, transport, specific components or all equipment from the supplier?
3. How many installed assets have been verified by model, revision, software version and site?
4. Which 4G, 5G and legacy services share the affected hardware or spectrum?
5. What passive, structural and power-system work is triggered by the new design?
6. What performance and energy results must a pilot achieve before national release?
7. Which costs are mandatory removal, and which are planned network modernisation?
8. What is the cost of missing the deadline, and what service risk is created by meeting it too quickly?

The most defensible replacement budget is not the one with the lowest number in the first presentation. It is the one in which every major assumption can be traced to an asset record, pilot result, contract term or measured site condition.

If your current estimate contains radios and labour but no migration, energy, spares, disposal or contingency, are you pricing a network replacement – or only the first purchase order?