Most significant software has moved from a one-time purchase to a recurring subscription over the last decade. The arguments for it are real and the costs are underdiscussed.
What actually changed
Under a perpetual licence you paid once for a specific version. It continued to work indefinitely. Upgrades were optional purchases.
Under a subscription you pay continuously for access. Stop paying and the software stops working, or drops to a restricted mode.
The important consequence is not the payment structure, it is that the ability to use the software is now conditional on an ongoing relationship with the vendor.
The genuine arguments for it
Worth stating properly rather than dismissing.
Continuous development. Predictable revenue funds ongoing work rather than a rush toward a paid upgrade every two years, and the software genuinely improves more steadily.
Lower entry cost, which makes expensive professional tools accessible to individuals and small businesses who could not fund a large purchase.
Cloud services that genuinely cost the vendor money continuously — storage, synchronisation, processing — cannot sensibly be funded by a one-time payment.
And security updates, which under the old model were frequently only provided for the current version, leaving people on older versions exposed.
What you lose
The costs, which are real and rarely enumerated by anybody selling the model.
Access to your own work. If the software is required to open your files and the subscription ends, the files are effectively inaccessible. This is the most serious issue and it applies to anybody with an archive.
Price certainty. Subscription prices rise, and the switching cost of an established workflow means the vendor has considerable pricing power.
The ability to stay on a version that works. Under a perpetual licence, an update that broke your workflow could simply be declined. Under a subscription, changes are pushed and staying behind is often not an option.
And the option to buy once and use for a decade, which for anybody with stable needs was frequently the cheapest arrangement by a wide margin.
The file format question
The thing I would look at hardest before committing to any tool.
If the software saves in an open, documented format that other tools can read, the risk of losing access is limited. You can leave and take your work.
If it saves in a proprietary format that only it can read, ending the subscription means losing access to everything you have made.
The practical mitigations are exporting to open formats periodically as a matter of routine, and checking before adoption whether a usable export exists.
This is not paranoia. Vendors change terms, discontinue products and go out of business, and it happens regularly enough that anybody with a long archive has probably experienced it.
Where subscription genuinely suits
Software with a real ongoing service component — synchronisation, collaboration, hosted processing.
Tools used intensively for a period and then not, where paying for months of use is cheaper than buying outright.
Anything where security updates matter and the alternative is running an unpatched old version.
And situations where the entry cost of a perpetual licence would be prohibitive.
Where it does not
Tools used occasionally, where an annual fee for a few hours of use is poor value.
Anything that is genuinely finished. Some software does not need continuous development, and paying continuously for maintenance of a stable product is paying for nothing.
And anything holding an archive you need long-term access to, unless the export path is genuinely reliable.
The alternatives that exist
Worth knowing about because the market is less uniform than it appears.
Some vendors offer perpetual licences alongside subscriptions, usually at a higher upfront price with a limited support window. These are frequently not prominently advertised and are available on request.
Open source alternatives exist for most categories, with the usual trade of capability and support against cost and control.
And some vendors have adopted licences where payment stops and the last version you paid for continues to work, which is a middle position and is the arrangement I now look for first.
Asking about that specifically, before adopting anything into a workflow, takes a few minutes and determines what happens in five years.
The organisational version
The same question at business scale, where the numbers are larger and the lock-in is worse.
Per-seat subscriptions scale with headcount, so a growing organisation faces a cost that rises with success and is difficult to reverse.
Migration away becomes progressively harder as data, integrations and habits accumulate.
The practical defences are negotiating multi-year price protection at the point of maximum leverage, which is before signing, and maintaining an export routine so that leaving remains possible in principle.
Neither is exciting and both are far cheaper than discovering the position after five years.
The bundle question
A pricing pattern worth noticing.
Vendors increasingly bundle several products into a single subscription at a price that makes the individual products look poor value.
That is good value if you use several and poor value if you use one, and the bundle is generally promoted more prominently than the single-product option.
The single-product tier frequently still exists and is harder to find, which is a deliberate design choice rather than an oversight.
Checking whether it exists before accepting the bundle takes a minute and is frequently worth a substantial annual difference.