Huawei: the future was part of the purchase

Related Signal: The terms of access can change examines how outside decisions reach the capabilities organisations are building, and what changes as AI enters their work.

In January 2019, Australian telecommunications company TPG Telecom (TPG) stopped building the mobile network it had begun rolling out two years earlier.

The equipment had not failed. Nor had an American semiconductor restriction cut off Huawei's supply chain; those controls came later.

TPG's problem was the future.

Its network design relied on what the company called a "simple upgrade path" from the Huawei equipment it was installing to Huawei 5G. Australia's 2018 security guidance prevented Huawei equipment from being used in 5G networks. TPG said that, with the expected upgrade path blocked and no suitable technical alternative available, continuing to invest in the network no longer made commercial sense. It abandoned the rollout and recorded an A$236.8 million impairment of associated spectrum and mobile-network assets for the year ended July 2019.¹

The decision was made in a wider commercial context: TPG was also pursuing a merger with Vodafone Hutchison Australia, and its account does not prove that no other strategy could have succeeded. But the mechanism is clear.

The value of the assets being installed depended partly on capability that had not yet been purchased.

Support, compatibility, product roadmaps and future upgrades rarely appear as separate physical assets. They can nevertheless be embedded in the economics of today's investment. Once TPG could no longer assume that today's Huawei network would lead into tomorrow's Huawei 5G network, the value of continuing to build changed.

The future was part of the purchase.

The strategic dependency was present before Huawei was placed on the US Entity List in May 2019, and before the semiconductor controls that followed. Operators had already built assumptions about future technology, vendor continuity and network architecture into investments being made in the present.

The later US actions exposed how much of that expected future depended on capabilities and relationships beyond Huawei itself.

A supplier can depend on a system its customer never contracted with

Huawei had substantial internal capability. Its HiSilicon business designed advanced semiconductors, and the company invested heavily in telecommunications hardware, software and research. But designing a chip was not the same thing as controlling the industrial system required to manufacture it.

In May 2019, the United States placed Huawei and dozens of affiliates on the Entity List, restricting specified transactions involving technology subject to US export controls.² Huawei had anticipated disruption, stockpiled components and substituted some US parts with its own designs. The United Kingdom's National Cyber Security Centre (NCSC) concluded that the immediate consequences could still be managed within the assurance arrangements then operating around Huawei equipment.³

The more consequential change arrived in May 2020. The United States amended its foreign-produced direct product rule to reach specified Huawei-related products made outside the United States using controlled US technology, software or production equipment. Further changes followed in August.⁴

The reach of the rule therefore did not stop at the US border or at the first tier of Huawei's suppliers. A chip could be designed by Huawei and fabricated abroad yet still be affected because of the tools and technology used to produce it. Semiconductor manufacturer TSMC later said it had stopped supplying Huawei by mid-September 2020.⁵

Huawei's telecommunications customers had not contracted with those semiconductor-tool companies, foundries or the US government. They could still depend on the route connecting them.

A supplier relationship can carry dependencies that sit several contracts away from the customer.

What happened in Britain shows why that matters. In January 2020, the UK had decided that Huawei could remain in parts of its 5G network subject to restrictions, market-share limits and a bespoke security-assurance regime. Six months later, the government decided that operators should stop buying new Huawei 5G equipment after the end of 2020 and remove Huawei equipment from UK 5G networks by the end of 2027.⁶

The change was not driven by the discovery that every installed Huawei component had suddenly become defective. The US rule had changed what Huawei might be able to manufacture next and, with it, what the UK believed it could continue to assure.

A decision several relationships upstream had reached infrastructure already being built downstream.

The rule reached the assurance machinery

The UK's Huawei arrangements were unusual. Since 2010, the Huawei Cyber Security Evaluation Centre had allowed UK authorities to examine Huawei products, software and engineering practices as part of a bespoke mitigation regime. The UK's willingness to tolerate Huawei in parts of critical telecommunications infrastructure therefore depended not simply on the equipment, but on the continuing ability to scrutinise it.³

The 2020 US rule disrupted that layer too.

The NCSC concluded that Huawei would have to redesign products and potentially develop new tools and supply arrangements under extreme time pressure. It was uncertain whether future equipment produced through those altered routes could be assured to the previous standard. More strikingly, the NCSC said the sanctions could prevent affected equipment, designs and binaries from being transferred to the evaluation centre while it remained inside Huawei UK, itself an entity affected by US controls.³

The export rule was therefore not only changing Huawei's manufacturing options. It was also reaching the UK's mechanism for deciding whether Huawei equipment could safely remain part of British networks.

The dependency had moved into assurance. An organisation may focus on whether a product continues to function, while a regulator or customer depends on an assurance process that establishes whether continued reliance remains acceptable. A capability can still operate while the evidence needed to support confidence in it is changing or disappearing.

Technical operation and assured usability are not the same state.

The NCSC's concern was explicitly prospective. Existing pre-sanction equipment could continue to sit inside the mitigation regime. The harder problem was future equipment, future upgrades and future support. It warned operators that long-term planning decisions being made in 2020 could leave them unable to replace or upgrade Huawei equipment later while still remaining inside the UK's security framework.

The future had become part of the risk assessment.

Removing one dependency could deepen another

The obvious response to a risky dependency is to remove it. The NCSC's own analysis shows why the answer was not that simple.

By 2020, three major suppliers served UK mobile access networks: Ericsson, Nokia and Huawei. In fixed access, the NCSC identified only Nokia and Huawei as major suppliers. Removing Huawei immediately from all new fixed-network equipment would therefore have pushed parts of the market towards a single remaining supplier.

The NCSC concluded that this reduction in supplier diversity could itself damage security and resilience. It recommended excluding post-sanction Huawei equipment from future 5G and data-transport use, while temporarily continuing to manage some Huawei fibre-access risk because the resilience consequences of moving immediately to one supplier could be worse.³

Reducing geopolitical dependency on one supplier can increase concentration dependency on another.

The trade-off also appeared in the treatment of equipment already installed. NCSC said the May 2020 rule did not at that point prevent Huawei from maintaining existing equipment, but it could not rule out future US action that might do so. It advised operators not to rely on Huawei for day-to-day managed services without an ability to change provider at short notice. For specialist third-line support, it recommended creating capability within the UK so that existing Huawei equipment could still be maintained if a future sanction interrupted Huawei's ability to provide critical fixes.³

A geopolitical intervention had therefore changed the capability the customer needed to possess for itself.

The question was no longer simply which supplier to buy from. It had become which functions could safely remain external, which needed an alternative, and which capabilities might have to be recreated domestically because the original supplier relationship could no longer be assumed.

Exit is not achieved by terminating a contract if the organisation still lacks the capability required to operate what remains.

Different jurisdictions reached different parts of the relationship

There was no single global event called the Huawei ban.

Different governments acted at different times, under different legal authorities, against different parts of the operating system.

Australia's 2018 security guidance changed the future upgrade path available to TPG before the later US semiconductor actions. Britain's 2020 decision distinguished between existing and future equipment and between mobile and fixed-network resilience. Germany's 2024 agreement with its major operators required Huawei and ZTE components to leave the 5G core by the end of 2026, with critical functions in access and transport network management systems to be replaced by the end of 2029.⁷

The United States faced a different problem: Huawei and ZTE equipment already embedded in smaller telecommunications networks. Its Secure and Trusted Communications Networks Reimbursement Program turned geopolitical unwinding into a financing, labour and execution problem. In June 2026, the Federal Communications Commission (FCC) reported final removal, replacement and disposal certifications for 53 of 126 Priority 1 applications, approximately 42 per cent; the programme required additional funding after estimated reimbursement costs substantially exceeded the amount initially appropriated.⁸

These are not simply different national attitudes to one Chinese supplier. They concern different objects of control: a future upgrade, eligibility for new equipment, assurance of future products, critical network functions, managed services, specialist maintenance and physical replacement of an installed estate.

The consequences therefore depend on where an organisation already is when the rule changes.

An operator that has only begun building can stop. An operator with sunk assets can face an impairment. A network with embedded equipment may need years of engineering work and public funding to unwind it. A regulator may tolerate an existing product but refuse to assure its successor. Another jurisdiction may permit access-network equipment while excluding core functions.

Geopolitics enters an operating model through the dependency that a particular rule can reach.

That makes global consistency an unsafe assumption. A multinational organisation can be using the same supplier and still occupy different strategic positions in different jurisdictions.

Huawei adapted. Its customers could not rewind

Huawei did not remain where the controls left it. Its reported group revenue fell from 891.4 billion Chinese yuan (CNY) in 2020 to CNY636.8 billion in 2021, then recovered to CNY880.9 billion by 2025. Huawei reported CNY192.3 billion of R&D spending in 2025, while more than 36 million devices were running HarmonyOS 5 or 6 by year-end.⁹

Those figures show substantial adaptation. They do not establish that Huawei recreated every capability, market or supply route available before the restrictions, and aggregate revenue is a poor measure of technological autonomy.

For customers, the more important point is that supplier recovery does not rewind decisions already taken around the dependency. A 2024 UK government-commissioned review of telecoms diversification found that Ericsson had replaced Huawei as the largest supplier of UK 5G radio-access-network (RAN) equipment, while Nokia and Ericsson had increased their shares as operators procured equipment to replace Huawei. It also noted that operators had made major financial investments in new 5G equipment and were therefore unlikely to undertake another large-scale procurement in the short term.¹⁰

By then, the customer position had changed in capital as well as policy. Money had been committed to replacement equipment, networks had been redesigned around different suppliers, and future procurement choices were constrained by those new investments. Huawei could recover as a company without recreating the option an operator had held before the transition began.

TPG's earlier impairment shows the same mechanism at the other end of the cycle: once the expected Huawei upgrade path disappeared, later developments could not retroactively restore the economics of a rollout that had already been abandoned.

A supplier can recover while the customer's former position remains gone.

Dependency analysis therefore cannot stop at the financial health or adaptability of the vendor. A resilient supplier may still leave the customer with a stranded architecture, a lost option or a capital programme already committed elsewhere.

The asset was larger than the equipment

Huawei is often presented as a story about whether governments should trust a Chinese telecommunications company. That question was real. It was also narrower than the dependency operators were actually managing.

Huawei depended on semiconductor designs, tools, fabrication and markets beyond its direct control. Operators depended on Huawei not only for equipment but for future supply, upgrades, support and compatibility. Governments depended on assurance mechanisms that themselves could be affected by export rules. Removing Huawei could reduce one form of geopolitical exposure while increasing supplier concentration. Replacing installed equipment created costs and execution requirements that took years to absorb. Huawei could adapt without restoring the choices its customers had already lost.

The network did not have to stop working for the value of the position to change.

What operators had bought was larger than the physical equipment. They had also bought assumptions about what would remain available around it.

The same mechanism extends beyond telecommunications. Long-lived technology decisions usually contain an expected future: software support, access to updates, compatibility with the next generation, specialist skills, regulatory acceptance, supplier capacity and the ability to assure the system as it changes. Those expectations can be economically material long before they appear as separate assets or explicit contractual obligations.

A procurement decision can therefore create exposure to decisions made several relationships away, in jurisdictions the organisation never considered part of the transaction.

The strategic question is not simply whether the supplier can deliver today. It is:

What future have we already purchased, and who else can change its terms?

The harder test is whether an outside decision could change the value of that future before the asset itself stops working.

Notes

Footnotes

1. TPG Telecom Limited, 2019 Annual Report, chairman’s letter, operating and financial review, and note 6 (pp. 45–46), year ended 31 July 2019. Management attributed the January cessation to the loss of its Huawei 5G upgrade path and the absence of a suitable technical alternative. The A$236.8 million total covered spectrum licences, mobile-network assets, related capitalised interest and other intangibles. Spectrum licences alone accounted for A$91.8 million. TPG was also pursuing a merger with Vodafone Hutchison Australia at the time. ↩

2. US Department of Commerce, Addition of Entities to the Entity List, published 21 May 2019, effective 16 May. This introduced licensing requirements for covered transactions, not a universal ban on every commercial relationship. ↩

3. NCSC, Summary of the NCSC analysis of May 2020 US sanction, 14 July 2020, especially paragraphs 14–17 (HCSEC), 21–24 (planning), 25–30 (maintenance), 31–40 (supplier diversity) and 41–42 (removal risks); A different future for telecoms in the UK, 14 July 2020. These are the contemporary UK technical authority's judgements, including uncertainty about future products and support. The assessment distinguished existing equipment from future supplies under changed production arrangements. ↩ ↩² ↩³ ↩⁴ ↩⁵

4. US Department of Commerce, May 2020 amendment, effective 15 May; August 2020 final rule, 85 FR 51596, effective 17 August. The case describes the historical mechanism at a high level, not the current treatment of a particular transaction. ↩

5. Reuters, TSMC says it has alerted US of potential China AI chip curbs violation, 22 October 2024, syndicated by The Indian Express. TSMC's statement said it had not supplied Huawei since mid-September 2020. This does not establish that no TSMC-made component later reached Huawei through another route. ↩

6. UK Government, Huawei to be removed from UK 5G networks by 2027, 14 July 2020; NCSC, The future of telecoms in the UK, 28 January 2020. The July timetable covered new 5G purchases after 2020 and removal by the end of 2027. ↩

7. Telefónica Deutschland, 2024 Annual Report, discussion of the July 2024 public-law contracts with the Federal Ministry of the Interior. Critical Huawei/ZTE components must leave 5G core networks by the end of 2026; critical functions of their 5G network-management systems in access and transport networks must be replaced by the end of 2029. These are different obligations, not a single deadline for every antenna. ↩

8. FCC, Secure and Trusted Communications Networks Reimbursement Program: Eighth Report, 15 June 2026. Recipients had filed final removal/replacement/disposal certifications for 53 of 126 Priority 1 applications; programme closeout was a separate stage, completed for 12 applications. The approximately 42% figure is application-based, covers eligible Huawei and ZTE equipment and services, and is neither Huawei's share nor a measure of physical equipment removed. The report also describes the initial funding shortfall and additional congressional funding. ↩

9. Huawei, 2025 Annual Report, financial highlights and research/innovation sections; 2022 Annual Report, comparative figures for 2020 and 2021. Values are rounded from Huawei-reported group results: revenue CNY891.4bn (2020), CNY636.8bn (2021) and CNY880.9bn (2025); R&D CNY192.3bn (2025). The 2025 report also records more than 36 million HarmonyOS 5/6 devices at year-end. Group revenue covers Huawei’s businesses collectively; it does not measure independence from foreign technology. ↩

10. Telecoms Supply Chain Diversification Advisory Council / DSIT, Telecoms supply chain diversification: independent report and recommendations, September 2024, paragraphs 36–37. The report records Ericsson's leading UK 5G RAN position, Nokia/Ericsson's replacement gains and recent investment constraining near-term large procurement. ↩