Post-Digital FuturesSignals · systems · strategyWork with Dean

Telecoms read-across · updated 3 October 2026

Why should
telecoms care?

Post-digital change is not confined to new computers. Processing is also moving into sensors, radio and switching ASICs, memory, optics and other network elements. These shifts could change where traffic is generated and processed, which capabilities are needed, how capex and opex move, and which services customers require. Most effects will come from hybrid systems, often exposed through familiar APIs and agentic workflows.

Executive answer

The opportunities span several technologies and timescales.

Telecom decision-makers should separate immediate operational requirements from investable adjacencies, bounded experiments and speculative developments.

01 · Act now

Establish the strategic baseline

  • Immediate commitments

    Begin post-quantum security migration and decide the organisation’s position on optical and AI infrastructure. These are current decisions, although they arise from different technology strategies.

  • Customised relevance map

    Assess each technology family against the organisation’s networks, products, customers, assets and supply chain. Give each one a likely timescale, confidence level and reason to care.

  • Ownership and governance

    Assign responsibility across strategy, technology, security, operations, investment, marketing, communications, policy, legal and innovation teams rather than leaving post-digital developments as an unowned CTO research watchlist.

  • Language and reputation baseline

    Review how the organisation uses ‘digital’ and emerging terms in its brand, corporate narrative and product language. Identify terminology that is useful, confusing, overhyped or likely to create ethical, legal or reputational concerns.

02 · Investigate now

Build evidence and external relationships

  • Scenario planning

    Develop three-, five- and ten-year scenarios covering possible adoption paths, supplier moves, new control points and implications for traffic topology, energy, skills and economics.

  • Targeted trials

    Test optimisation, edge inference, in-memory, neuromorphic, photonic or RF-domain processing only where there is a defined constraint, a strong conventional baseline and a measurable outcome.

  • Industry participation

    Select the most relevant industry bodies, standards groups, conferences and technical consortia. Participate to learn and influence, without treating membership as evidence that a market exists.

  • Research partnerships

    Build relationships with universities and research teams working on relevant workloads, materials and systems. Use sponsored projects, testbeds or doctoral research where they provide access to scarce expertise.

03 · Prepare selectively

Create strategic and investment options

  • Government-funded programmes

    Consider research grants, sovereign-technology programmes and public-private consortia where they fund useful capability, test infrastructure or supply-chain access rather than a publicity-led demonstration.

  • Venture and corporate development

    Evaluate startup partnerships, minority investments, corporate venture positions and possible acquisitions where they create learning, access or a defensible control point.

  • Internal capability

    Develop the skills, data, test facilities, assurance methods and architecture needed to evaluate specialist hardware and integrate it behind familiar software and operational tools.

  • Commercial options

    Explore possible retail and wholesale roles in connectivity, hosting, assurance, integration, managed optimisation and research infrastructure before committing to a generic post-digital product portfolio.

04 · Watch, do not plan around

Maintain an informed long-term position

  • Structured horizon scanning

    Track frontier work in spintronics, magnonics, physical and ionic systems, reservoir computing, biological computing and other category-resistant approaches.

  • Decision triggers

    Define the evidence that would move a technology into active investigation, such as independent replication, credible system benchmarks, manufacturing progress or a relevant commercial workload.

  • Policy, regulation and capital signals

    Monitor government funding, venture activity, university spin-outs, specialist hiring, supply-chain investment and emerging rules for spectrum, sensing, AI, security, sovereignty and biological systems. Treat them as signals rather than proof of adoption.

  • Avoid premature commitments

    Do not use further-future mechanisms as assumptions in network architecture, procurement, capex, opex or service roadmaps until the evidence changes materially.

The central read-across

Focus on where processing takes place, the assets it uses and the guarantees it requires from the network.

Information may move among devices, access sites, metro edges, cores, datacentres and remote specialist processors in new patterns. Raw streams may become compact events, while model distribution, telemetry and control loops follow different paths. Direction, locality, timing and sovereignty may matter more than aggregate volume.

Networks and computers are also becoming harder to separate. AI clusters increasingly resemble specialised optical networks, while sensors, radio ASICs, switching silicon, optical components and memory can perform selected calculations. This could change both internal network capabilities and the adjacent services operators are asked to provide.

Seven impact areas

Where architecture, economics, services and positioning may change.

These are questions to investigate and track, not quantified forecasts. Each area links to the evidence and primer needed to examine the claim.

01

Traffic location, direction and topology

Start by mapping where information originates, where it is processed and which network domains it crosses.

Why it could matter

  • Spatial dataflow, analogue in-memory learning and digital processing in memory could make more analysis viable inside devices or small edge installations. Traffic then shifts towards alerts, selected evidence and model management, with raw data retained locally where useful.
  • Local adaptation adds another traffic pattern: devices may learn separately, while fleet coordination, validation and software updates still cross the network. Operators and integrators need to decide what stays local, what is shared and how the resulting behaviour is audited.
  • Onboard satellite processing could alter downlink demand and scheduling. The quantum photonic experiment in orbit is an early payload-engineering result; useful sensing and communications applications still need to be developed.
  • Map these changes separately for premises, access sites, metro facilities and remote clouds. A reduction in one raw-data stream can coexist with more distributed processing locations and management traffic.

Questions before acting

  • Where does the workload originate, where is it processed and where does its output terminate?
  • Is the traffic raw data, features, model updates, control information or access to a remote specialist processor?
  • Which parts traverse access, aggregation, metro, core, datacentre or international networks, and which must remain local or sovereign?
  • Could satellite, home-gateway or in-building processing create a new path that bypasses, complements or changes a terrestrial network domain?
Related evidence and primerSignal 061 · Spatial dataflow at the endpoint →Signal 063 · Learning within memory →Signal 079 · Quantum photonics in orbit →Signal 080 · Processing in memory within a server →
02

Specialist assets and network architecture

Some post-digital systems may use familiar connectivity, while others require new optical, edge, timing, hosting, security or assurance capabilities.

Why it could matter

  • Heterogeneous computing has physical placement constraints. Q.ANT presents cloud and on-premises options; Extropic’s proposed hybrid system exchanges intermediate results between different processor types. Dataset size, utilisation and transfer time influence where each component belongs.
  • Backline reports a local quantum-control loop below three microseconds in its tested configuration. Such feedback favours close placement and predictable interconnects. External cloud access serves job submission and other less time-sensitive work, with a different latency budget.
  • Fujitsu’s proposed modular diamond-spin architecture depends on optical interfaces. Wavelength conversion, optical loss, entanglement rate and memory lifetime would constrain connections beyond the local modules, influencing future facility layout.
  • Cryogenic logic could move processing across the boundary between room-temperature equipment and cold quantum devices or sensors. That changes wiring, power and interface requirements even where the public network sees no new traffic.

Questions before acting

  • Which assets are genuinely new, and which are upgrades or extensions of existing fibre, datacentre, edge, timing and operational infrastructure?
  • Does the capability need to sit inside operator infrastructure, customer equipment, a vendor platform or a specialist cloud?
  • Which new skills, test methods, maintenance processes and service-assurance metrics are required?
  • Could new device categories emerge around sensing, local AI, quantum security or satellite access, and who would certify, manage and support them?
Related evidence and primerSignal 062 · Photonic cloud and on-premises options →Signal 066 · Hybrid processor placement →Signal 067 · Computing within cryogenic equipment →Signal 073 · Optical connections between quantum modules →Signal 076 · Fast local quantum control →
03

Capex, opex and lifecycle economics

Post-digital technologies may lower some compute, energy or transport costs while adding expenditure elsewhere in the system. The important issue is where cost moves.

Why it could matter

  • Processing beside a sensor or within memory could change the minimum viable size of an edge installation. Evaluate the complete service, including utilisation, model loading, field support and the cost of managing more sites.
  • Hybrid processors can move expenditure into local interconnect, conversion and orchestration. Extropic’s energy projections exclude final dense output calculation and inter-processor data movement, making both important to a deployment assessment.
  • Quantum hardware simplification could reduce repeated control and installation overheads. Alice & Bob’s coupler experiment is chiefly relevant within a machine; any eventual economic benefit depends on the complete wiring and control system.
  • Optical integration and lower-loss fabrication may relax other system requirements, while new timing sources add power and maintenance costs. Component savings need to be evaluated against the capital and operating costs of the resulting architecture.

Questions before acting

  • Does a claimed saving survive at complete-system level after packaging, conversion, control, cooling, resilience and staffing are included?
  • Is cost removed, deferred or transferred to a cloud platform, equipment supplier, customer device or another network domain?
  • How do upgradeability, asset life, utilisation, vendor concentration and exit costs affect the investment case?
Related evidence and primerSignal 066 · What the energy projection excludes →Signal 071 · Simpler quantum control hardware →Signal 075 · Optical fabrication and loss →Signal 078 · Local timing economics →Signal 080 · Economics of smaller inference sites →
04

Internal network capability and performance

The nearer operational value may come from improving how networks are planned, sensed, optimised, maintained and assured rather than changing the external service itself.

Why it could matter

  • Neuromorphic sensing could support infrastructure monitoring, interference classification and local alerts. The UK roadmap and BrainChip evaluation hardware provide routes to investigate these applications in realistic environments, including buildings, remote equipment and private wireless networks.
  • On-device learning needs operational controls: establish when adaptation is allowed, how it is validated and how a device can return to a known state. These responsibilities connect edge hardware with fleet management and assurance software.
  • The experimental optical transmission system in Signal 069 moves some signal-processing work into intermediate photonic nodes. This changes where a network function runs and where it needs configuration and calibration, below packet routing and software orchestration.
  • Compact timing sources could extend holdover at selected sites when an external reference fails. Assess drift across temperature and ageing, recovery behaviour and compatibility with terrestrial timing distribution alongside existing oscillators.

Questions before acting

  • Was the system used in production, or only for a planning exercise or trial?
  • Which network KPI, operational decision or cost would improve, and what is the strongest conventional baseline?
  • Can the capability be monitored, overridden, updated and supported within existing operational and assurance processes?
  • Where would it connect to OSS, BSS, service orchestration, assurance, AI-RAN or network-digital-twin workflows, and which system remains the source of operational truth?
  • When a proposal is described as ‘AI-native’, which decisions, models, data loops and network functions are actually different?
  • Does sensing or adaptive spectrum use create new privacy, safety, licensing, interference or lawful-access obligations?
Related evidence and primerSignal 063 · Assuring local learning →Signal 064 · Neuromorphic testbeds and adoption →Signal 065 · Evaluating event-based sensing →Signal 069 · Optical processing along the fibre path →Signal 078 · Timing resilience and holdover →
05

Adjacent services and customer guarantees

Commercial opportunities may emerge from the network capabilities required around post-digital systems, even when operators do not supply the processor itself.

Why it could matter

  • Quantum compute access needs an end-to-end service description that includes queueing, preparation and network delay as well as reliable computation. CUDA-Q Logical and QUOPS address machine performance; providers must add the customer-facing service measurements.
  • Logical-qubit counts need context about encoding, circuit depth and usable output. Infleqtion’s experiment illustrates why service buyers will need clear explanations of error protection and repeatability when comparing processors.
  • Eurofiber’s metropolitan testbed points towards possible managed quantum links. Candidate measures include entanglement delivery rate at a stated fidelity, availability and the impact of route changes or maintenance. These would supplement conventional fibre metrics.
  • Specialist access, local integration and link assurance may become distinct services. Distributed sensing could share some quantum-network requirements with computing; resilient timing provides a separate opportunity in existing networks and private installations.

Questions before acting

  • What customer problem requires a network guarantee or asset that ordinary connectivity does not provide?
  • Can the capability be measured, assured and expressed through an SLA, API, managed service or contractual commitment?
  • Does the operator control a defensible asset or capability, or would most value accrue to a hyperscaler, equipment vendor or specialist integrator?
  • Is the better route a retail service, a wholesale input to another provider, a neutral-host model, or an internal capability that improves an existing offer?
  • Could regulation turn provenance, locality, spectrum coexistence, model assurance, sensing consent or device certification into a service requirement?
Related evidence and primerSignal 070 · Quantum performance and service delivery →Signal 072 · Logical qubits and usable output →Signal 076 · Local control versus remote access →Signal 077 · Metro quantum-link engineering →Signal 078 · Resilient timing →
06

Supply chains, sovereignty and strategic control

Telecom relevance extends beyond networks and traffic to control of foundries, materials, packaging, specialist software, skills, test capacity and intellectual property.

Why it could matter

  • The US CHIPS awards to GlobalFoundries and Quantinuum put quantum foundries, integration and specialised optical components into the infrastructure-sovereignty discussion. Access to manufacturing capability can matter as much as the location of a computing service.
  • Xanadu and ASML’s work on optical loss highlights fabrication quality as a constraint on the size and layout of photonic quantum machines. Network component makers may find adjacent opportunities in processes, packaging and test equipment.
  • Some capabilities can serve telecom optics, sensing and quantum systems, but materials and operating wavelengths affect how directly a process transfers. Evaluate the actual manufacturing overlap before assuming a shared market.
  • Memory suppliers are also shaping compute architecture. Procurement and strategy teams should track their processing-in-memory roadmaps alongside processor vendors, including software support, integration requirements and supply continuity.

Questions before acting

  • Which control points are defensible for an operator or network vendor?
  • Which dependencies create unacceptable concentration, geopolitical, licensing, export-control or support risks?
  • Which capabilities should be owned, partnered, funded through research, accessed through venture investment, procured as services or simply monitored?
  • Which regulatory questions should be influenced through standards bodies, consultations, government programmes and industry alliances before rules solidify?
Related evidence and primerSignal 073 · Diamond-spin photonic integration →Signal 074 · CHIPS quantum-manufacturing awards →Signal 075 · Lithography and optical losses →Signal 080 · Memory suppliers influence compute →
07

Corporate narrative, branding and organisational alignment

Post-digital change is also a communications and positioning issue. Telecom companies and vendors need language that is credible to customers, investors, policymakers, employees and partners before every technology has a mature product category.

Why it could matter

  • Explain the architectural reason for inclusion. Efficient Computer’s spatial dataflow remains digital, but changes how operations are placed and scheduled. Its relevance to this observatory is the post-von-Neumann execution model and possible effect on compute placement.
  • Distinguish near-term evaluation from longer-term applications. Development hardware and software can support practical trials; cryogenic integration and modular quantum links have different engineering dependencies and adoption paths.
  • Keep room for foundational research with limited telecom relevance. The DNA-assembly computer belongs in the wider technology watch because it computes through molecular interactions; its clearest connection is to chemistry and biological experiments.
  • Give strategy, technical and communications teams a shared account of what changes physically, what has been demonstrated and what remains a possible application. That produces a more useful narrative than treating every new chip or efficiency claim as the same trend.

Questions before acting

  • Does existing ‘digital’ positioning still describe the organisation’s future role, or does it need to broaden without discarding a useful and recognised term prematurely?
  • Which terms should be adopted, carefully explained, qualified or avoided for different audiences and markets?
  • Can the organisation explain how its OSS, BSS, AI-RAN and wholesale strategies connect to post-digital developments without implying that every existing initiative is part of the same trend?
  • Who approves technical claims and ensures that product, policy, legal, investor and corporate narratives remain consistent?
  • Where could commissioned analysis, executive workshops, messaging development or external thought leadership create a defensible point of view?
Related evidence and primerSignal 061 · Spatial dataflow and the scope of post-digital →Signal 067 · Cryogenic integration →Signal 068 · Molecular computation →Post-digital technology primers →

Implications by role

The answer depends on where you sit.

A technology can matter to the telecom sector without becoming an operator service or appearing directly in a public network.

Operators

Map where workloads sit and which network domains they cross, including satellite, in-building and residential environments. Assess implications for assets, capex, opex, OSS/BSS, AI-RAN, wholesale models, spectrum coexistence and possible service guarantees. Avoid a generic product portfolio without a customer problem and defensible capability.

Network and software vendors

Identify where post-digital components could improve radios, transport, edge platforms, OSS or BSS. Build the assurance, orchestration, calibration and fallback layers that make heterogeneous hardware operationally supportable, and be precise about what ‘AI-native’ changes.

Cloud, datacentre and infrastructure providers

Treat workload placement, optical topology, power, cooling, locality, timing and connectivity as one system. Clarify who owns and assures each control point across cloud, edge and network domains, and whether access is sold directly or through wholesale partners.

Investors and corporate development

Distinguish scientific progress from a manufacturable component and a component from a supportable company. Use partnerships and minority investments to build options without mistaking technology scouting for a near-term revenue forecast.

Policymakers and research funders

Look beyond headline compute performance to foundries, materials, packaging, test capacity, skills, standards and supply-chain resilience. Anticipate questions around spectrum coexistence, sensing privacy, device approval, AI assurance and sovereignty, while funding shared capabilities and credible application pathways.

Universities and research groups

Work with operators and vendors on realistic datasets, field constraints, assurance requirements and conventional baselines. Telecom relevance should be demonstrated through a system problem, not added as a speculative final paragraph.

Marketing, communications, policy and legal teams

Work with the CTO’s office early to understand which terms are credible, which claims need qualification and which developments may create regulatory, ethical or reputational sensitivity. Use evidence-led thought leadership to shape categories without presenting speculation as a product roadmap.

Apply the analysis

Four ways to use this work.

The objective is to turn emerging technologies into clearer decisions and a credible external narrative for a particular organisation, market or audience.

01 · Scenario planning and commissioned analysis

Build scenarios and a customised relevance map.

Develop three-, five- and ten-year scenarios, test assumptions and map possible consequences for traffic topology, assets, capex and opex, internal capabilities, services, sovereignty, regulation and strategic positioning.

Discuss a scenario-planning project →
02 · Executive workshop

Turn weak signals into decisions.

Bring strategy, network, product, security, regulatory, innovation and corporate-development teams together to test assumptions, prioritise experiments and define decision triggers.

Discuss a workshop →
03 · Speaking and briefings

Explain the field without the hype.

Use a keynote, leadership briefing or customer-event session to connect post-digital developments with optical, radio, satellite, edge, device, service and policy strategies.

Discuss a speaking engagement →
04 · Thought leadership and messaging

Develop a credible post-digital narrative.

Audit terminology including ‘AI-native’ and ‘AI-RAN’, develop messaging and content themes, align the CTO’s office with marketing, communications, policy and legal teams, and connect technical change with OSS/BSS, wholesale and service strategies.

Discuss thought leadership support →
Explore Dean Bubley’s wider advisory and speaking work ↗

Reality check

What telecom leaders should not assume.

Aggregate traffic is the wrong starting point.

Where traffic originates, terminates and is processed, which network domains it crosses, and what latency, timing, locality or sovereignty it requires may matter more than total volume.

A device benchmark is not a system advantage.

Lasers, converters, memory, calibration, cooling, packaging, communications and software can erase an impressive laboratory efficiency figure.

Lower unit cost does not guarantee lower network expenditure.

Savings in compute, transport or energy can be offset by edge deployment, optical upgrades, specialist testing, integration, scarce skills, resilience and supplier concentration.

A telecom use case does not mean a telecom buyer.

The accelerator may be bought by a network-equipment vendor, cloud provider, device maker or semiconductor company and remain invisible to the operator.

A trial is not an operational deployment.

Planning exercises, demonstrations and limited proofs of concept should not be described as live network control or evidence of scalable economics.

Sovereignty is more than data location.

Control may depend on foundries, packaging, materials, intellectual property, export rules, specialist software, skills and operational responsibility as well as where information is stored.

Further-future research still deserves attention.

Early spintronic, reservoir, ionic and biological work can reveal new mechanisms and strategic options. It belongs on a watchlist, not in a procurement forecast.

Living analysis

Recent revisions

This section updates the strategic interpretation when several signals point in the same direction.

Compute placement has several distinct network consequences

The complete October edition, Signals 061–080, adds evidence on local learning, heterogeneous racks, fast quantum feedback and optical modules. Each creates a different connectivity requirement. The revised seven impact areas now include all twenty signals, while retaining the limited telecom relevance of foundational molecular work.

Signal 063 · Local learning →Signal 066 · Hybrid racks →Signal 076 · Quantum control →Signal 079 · Satellite payloads →

Manufacturing changes the feasible networked system

Quantum foundry investment and photonic lithography work connect telecom-adjacent supply chains with the physical layout of future machines. Lower loss and better integration could affect how modules connect, while shared manufacturing capabilities may give component suppliers several routes to market.

Signal 074 · Quantum foundries →Signal 075 · Lower-loss fabrication →

Traffic topology and network capability matter more than volume

The telecom impact depends on where information is generated, processed and consumed; which access, edge, core and datacentre domains it crosses; and what latency, locality, assurance or sovereignty the workload requires. Aggregate traffic is only one possible consequence.

Signal 032 →Signal 034 →

The compute/network boundary is becoming the main story

July’s photonic work reinforces the view that telecom components are becoming computing components. Optical control, packaging and assurance may be more defensible adjacencies than photonic AI chips themselves.

Signal 027 →Signal 031 →

Quantum enters operations through a side door

AT&T’s D-Wave work matters because it is being placed inside familiar operational and agentic tools. The likely model is hybrid optimisation as a capability, while evidence of quantum advantage remains absent.

Signal 021 →

Security migration is not a quantum internet

The first run was dominated by post-quantum cryptography, QKD and key management. Operators should act on cryptographic migration without allowing quantum-network language to inflate near-term infrastructure expectations.

Signal 002 →Signal 006 →

Continue the exploration

Map possible futures before deciding what matters.

Most telecom organisations do not yet have a clearly defined post-digital problem, use case or strategic position. Use the primers and signals to build three-, five- and ten-year scenarios, expose assumptions, identify uncertainties and decide which developments deserve deeper attention.

Understand the background: read the primersTrack the signals ↗