Living study Last updated Revision Next review

Announced is not deployed

Mobility coverage treats the two as the same thing. This study separates them and shows its working. Filter the technology stack by what is actually in revenue service and see how much of the conversation survives.

60technologies classified by evidence
9conditions recurring across 16 regions
10gaps tested, 9 already occupied
8ways AI changes the question

The evidence filter

What is actually running

Every technology below carries an evidence class. Deployed means in commercial revenue service with disclosed volume. Announced means a funded programme with a target date, which is not the same thing and is usually reported as though it were.

Mode by mode

Where the evidence is strong, and where it is not

Each mode gets the same three questions. The third is the commercially interesting one.

    The core argument

    Conditions, not countries

    A handful of operating conditions recur across markets that share no border, climate policy or supply chain. Build against a condition and you address five markets at once. Select any row or column.

    Scroll the grid sideways to reach every region.

    Condition

    The layer that is not a layer

    AI is not one item in the mobility stack

    AI is easy to list alongside propulsion and signalling as though it were a comparable component, and this study did exactly that until now. It is not comparable. It is changing what a vehicle is, how safety is proven, how operations are staffed, and how travel is bought. Eight effects, each labelled by how well it is supported, and the third is the one nobody has solved.

    Where technologies merge

    The value is in the joins

    Vendors compete on components. The defensible positions sit at the seams between them. These are propositions rather than findings, so each carries a confidence label: observed, asserted or hypothesis. Select a play to see the merge and what follows from it.

    Thinking loud

    other technologies that will touch mobility

    Beyond AI, and beyond the obvious. Materials, physics based sensing, propulsion outside road, and the institutional layer that most technology studies leave out entirely. Some are running now and barely discussed. Two are included specifically because the honest read is that they matter less than the coverage suggests. Filter by when each one actually bites.

    The blind spot

    Maintenance takes most of the lifecycle cost and least of the attention

    There is a structural reason for it. Capital procurement is visible, event driven, politically rewarding and vendor supported. Maintenance is continuous, unglamorous, distributed across thousands of small decisions, and no supplier's interest is served by making it cheaper. Twenty-one fractures, grouped by what is actually broken, each labelled observed where it is documented practice and asserted where it is judgement. Select any one for why it persists and what fixes it. Nothing here is a maintenance instruction: intervals, procedures and safety cases belong to the operator and its regulator.

    Reading the labels

    Every claim says how much it is worth

    Filter this sample down to the documented claims and watch how much survives. The rest is judgement, and it is labelled as such on every line.

    The incumbent check

    We checked who is already doing it

    A gap is only a gap if nobody occupies it, so each one is tested against companies already operating. Nine of these ten turn out to be occupied, and what remains is narrower and more specific than the gap first appears. A company is named only where a public source is linked alongside it. Where no citable source was obtained, the entry describes the category and names nobody. Nothing here evaluates any company as a supplier, counterparty or investment.

    Obligations

    Only the obligations we have actually verified

    Obligations are cumulative across the vehicle, the operator, the infrastructure, the data and the AI. On a compliance topic an unverified entry is worse than a missing one, because a reader might act on it. So this section carries only obligations traced to an official source, and says plainly what it does not cover.

    This is a scoping aid and not legal advice. Take qualified advice in each jurisdiction before relying on any of it.

    What this section does not cover

    European Union obligations only, and not all of them. Nothing here addresses the United States, China, India, the United Arab Emirates or any other jurisdiction. Within Europe, the rail interoperability and safety instruments, the accessibility act, data protection, the multimodal data access rules and the international maritime framework are all absent. None of them could be traced to an official source for this edition, so none of them appear. An unverified obligation is worse than a missing one, because a reader might act on it.

    Treat the absence as an absence. It does not mean no obligation exists, and a European design will not transfer without local advice.

    The register

    Every number, including the ones that failed

    Every specific figure used anywhere in this study appears here with its source, tier, date and verification status. That is the rule: if a number cannot be traced, it does not get published as a number. One rejected claim is included, because a study that never finds itself wrong has not been checked.

    Figure as usedSourceTierAs atStatus

    Before you commit

    Test an idea in five questions

    Every proposition in this study can be tested for a fraction of what it costs to build. Five questions, one minute, and the readout names only the checks you failed and why each one matters.

    One answer in each pair is the strong position. Answer as the sceptic rather than the sponsor, and treat anything short of a confident yes as a no. Nothing is scored, stored or sent anywhere. Passing all five does not validate an idea; it means the premise survived the cheap checks. The propositions in this study would have passed most of these, and eight of ten still fell over on question three.

    Revision record

    What has changed, and what will change

    This study is maintained rather than published once. Corrections will appear here rather than quietly disappearing, including the occasions when checking the work overturns it. This is the first issue, so the record begins here.

      Figures carry their own verification date in the register. Deployment data in fast moving segments ages within weeks, so the review cadence is deliberately short.

      Updates to this study

      Be told when this study changes

      This is a maintained study, not a fixed document. Figures carry a verification date, deployment data in fast moving segments ages within weeks, and corrections are published in the revision record rather than quietly applied. Leave an address and we will tell you when it is revised, and send the full report behind it.

      Nothing here is gated. The study stays open whether you leave an address or not.

      One address, no password, no drip sequence, and nothing about other subjects. Unsubscribe is a single click in any message we send.

      Contact us

      Take this further

      A study describes a market. Applying it to a specific fleet, corridor, depot or entry plan is a different piece of work, and that is what we do: market entry across the Middle East and Africa, operations and maintenance strategy, and the regulatory scoping that usually sits on the critical path.

      Corrections are equally welcome. This study records what it is unsure about and publishes its revisions in the open, so evidence that changes a conclusion improves the next issue rather than embarrassing this one.