← BlogAI in the plant

Technology has a lifespan: the hidden cost of late decisions

Time to value belongs in every technology decision. How to weigh the cost of delay against the risk of acting early, and when waiting is the wiser choice.

A technology decision can be right and still lose much of its value if it is carried out late. This article explains where that cost comes from, how to estimate it with simple arithmetic, and when waiting really is the better choice.

Technology does not wait for organizations

In many organizations and industries, choosing and adopting a new technology has become a long process. Meetings, feasibility studies, vendor comparisons and approval rounds can take months, sometimes years.

These steps exist to reduce risk and to pick the right solution. But one variable is often overlooked: time. Organizations ask which technology is better and what it costs. They rarely ask:

If we adopt this technology a year later, how much of its potential value will we lose?

This is not only a theoretical worry. In its official review of digital spend (March 2025), the UK government found that current processes are too complex for many digital initiatives and experimental technologies, which delays decisions and slows down the delivery of services.

A technology is not just a tool or a technical asset. It is an opportunity to create value, and the economic and competitive conditions around that opportunity change over time.

A technology's working life is not the same as its value life

A technology can keep working for years while its economic or competitive advantage changes much sooner. Three lifespans need to be kept apart:

  • Technical life: the period in which the technology can still be used and maintained.
  • Economic life: the period in which its benefits justify its costs.
  • Competitive life: the period in which it gives a meaningful advantage over competitors.

These periods do not necessarily coincide. The technology value window is the period, or periods, in which using a technology creates an opportunity for value, given its technical maturity, the market and the organization's readiness.

In this article the term is a proposed analytical framework, not an empirically validated model. Nor does the value of a technology always fall over time: lower costs and greater maturity can make later adoption more valuable.

So the problem is not only that technology ages. The value of the opportunity to use it changes over time.

The decision paradox: when reducing risk becomes a risk

To avoid the wrong choice, organizations build review and approval procedures. They are necessary. But when the cost of delay is not weighed alongside the risk of a wrong choice, the procedures themselves can become the obstacle.

Between review and approval, new solutions appear or the project's economic assumptions change. The review is repeated and the decision cycle starts again. The result can be a large spend of time and resources without matching operational value.

Research in strategic management also links the speed of decisions and of implementation to company performance, although the link depends on the setting and on how it is measured. A meta-analysis by Kolev and colleagues (2025), which combines the results of 127 studies, reports an overall positive effect of 0.21 for decision speed on firm performance, 0.22 for implementation speed and 0.10 for the speed of response to competitive challenges. On its own, this does not prove cause and effect, and it does not measure the cost of delay in adopting technology directly.

A lot of activity around transformation is not the same as transformation.

The cost of delay is value that never shows up on an invoice

The cost of delay is the economic consequence of value arriving later. The Project Management Institute (PMI) describes it as what a delay in realizing value costs an organization: revenue and opportunities lost, higher risk, and lost standing with customers.

Suppose a smart monitoring system for industrial equipment produces a fixed monthly benefit once it is running. Over a 12-month horizon, we compare two scenarios: two months of preparation with operation from the start of month 3, or eight months of preparation with operation from the start of month 9.

What six months of delay does to the first year's benefit

MeasureOn timeDelayed
Evaluation and deployment2 months8 months
Start of operationStart of month 3Start of month 9
Months of benefit in year one10 months4 months
Year-one benefit, relative to on time100%40%

This is a hypothetical example. The monthly benefit and the costs are the same in both scenarios, and discounting, a gradual ramp-up and changing market conditions are left out.

In this example, six months of delay removes 60% of the benefit available in the first year. If the net operating benefit were 500 million tomans a month (the toman is the unit Iranians use every day for prices, equal to 10 rials), the gap would be 3 billion tomans. That figure is the difference in benefits over a fixed horizon, not a full return-on-investment calculation, and it does not prove that the benefit is lost for good.

Delay does not necessarily shorten a technology's technical life. But it can shrink the organization's share of its value window.

Delay can also prolong downtime, rework, wasted resources and inefficient processes. And operational learning starts later, an advantage that rarely appears in a project's direct calculations.

When technology moves faster than the organization

AI is a clear case of fast-changing conditions. According to Stanford University's AI Index 2025, the inference cost (the cost of running a model to get an answer) of a system performing at the level of GPT-3.5, the model behind the first version of ChatGPT, fell more than 280-fold between November 2022 and October 2024. The figure applies to that level of performance, not to every model or to the full cost of deploying AI in an organization.

In industry, too, the value of a technology appears only when it moves from selection to operation. In January 2025, the World Economic Forum added Beijing Shougang Cold Rolling to its Global Lighthouse Network, a group of industrial sites that use digital technology at scale. The Forum reports that the plant deployed 67 Fourth Industrial Revolution use cases, 61% of them based on AI, cutting product defects by 35% and customer complaints by 55% and improving production-line efficiency by 21.2%. The example shows operational value being realized; it does not measure the cost of delay directly.

Seen this way, three time scales have to be watched together: how fast the technology changes, how long the organization takes to decide, and how long deployment takes to deliver value. An organization can decide quickly and still arrive late because integration, training or changes to its processes are slow.

Faster is not always better

Acting early can mean high cost, immature technology, dependence on one supplier and operational risk. In safety-critical industries, the cost of haste is sometimes higher than the cost of delay.

Daniel Forbes (2005), for example, studied 98 internet start-ups and, in exploratory analyses, found that those making faster decisions were more likely to close. The finding is limited to that sample and setting, and should not become a general rule that speed does harm.

Real options theory, which treats an investment as a right that can be used now or later, also shows that under uncertainty it can pay to keep the option of investing later and to gather more information first. Rita McGrath (1997) extended this logic to technology investments and the timing of acting on them. So the choice is not simply speed against accuracy. It is weighing the cost of delay, the risk of acting early and the possible value of waiting, all at once.

A first guide to when to act

Cost of delay / risk of actingLow risk of actingHigh risk of acting
High cost of delayAct now: start deployment and measure the benefitsControlled test: a limited pilot project and a staged decision
Low cost of delayPlanned action: implement according to capacity and priorityWait and watch: gather information and set a review date

The matrix is this article's proposed framework. Its categories are relative and do not replace an economic analysis or safety requirements. The value of waiting, whether the decision can be reversed, and whether a pilot project (a limited but real run on one specific problem) can be carried out safely also need to be checked.

The matrix is for choosing the next step. Each path needs a clear expected benefit, a criterion for continuing or stopping, and a review date. Staged decisions, controlled tests and a flexible architecture let an organization measure benefits early while keeping the freedom to learn, upgrade or replace the technology.

Time belongs in every technology decision

Technology assessments usually weigh capabilities, cost and risk. Time to value belongs in the economic and management assessment as well.

A decision can be technically right and fully compliant with procedure, yet be carried out so late that much of its value window has passed. Equally, a hasty decision does not necessarily produce more value. Before approving a technology investment, one question helps:

If we postpone this decision by six months, what do we gain, and what value do we lose?

An organization's lasting advantage lies not only in choosing the right technology, but in deciding on time, turning the choice into real benefit, and adapting continuously to change.

Technology does not wait for organizations to decide. An opportunity that creates value today may not offer the same value tomorrow.

Vakav's view

At Vakav we start projects with the same logic. The VAI platform is installed on a server inside the plant, and work can begin before any live connection: from Historian exports (the control system's data archive), a model is built on the unit's own historical data and the expected benefit is measured before a large investment. If the result is convincing, the same model runs on live data through a direct connection to the distributed control system (DCS), and its results appear on the control-room display. The operator decides whether a recommendation is written back to the DCS.

This is the controlled test in the matrix above: a small step, a clear criterion for continuing or stopping, and data that stays inside the plant. We described one such path in predicting sponge-iron quality before the lab result.

If a technology decision has been waiting on your desk for a while, see the VAI page or talk to us about a pilot project.