Markets rarely move in straight lines, and AI is no exception, but there is a useful reference point we can look back on in order to plan how we move forward. The conditions aren’t identical but the dot com era illustrates how periods of rapid technological optimism tend to evolve.
There are some stand out examples, like Pets.com which gambled millions on pet-food ecommerce before consumers had fully changed habits – and disappeared quickly. On the other hand there are companies like Amazon that sought to solve real problems, even if they did so imperfectly, endured and went on to define the next economic cycle.
The AI market is much more concentrated than the early internet ever was, and the structural warning signs are real – this concentration increases systemic risk. When valuations run far ahead of fundamentals in a narrow part of the stack, any slowdown reverberates widely. The AI system is brittle so how can we ensure its longevity and continued value?
Measuring AI’s impact on long-term enterprise value
In order to make that shift towards long-term value, organisations must move away from the “instant gratification mindset” that prioritises marginal gains in personal efficiency or faster content creation. We must move past speculative growth to identify what value survives a correction, adopting what I call “cathedral thinking” which is an outlook focused on building systems designed for longevity and resilience rather than just the next launch.
Internally, this requires a change in intent and discipline. Instead of treating AI as a glorified spreadsheet or a tool for short-term visibility, leadership must optimise for infrastructure-level utility. Success belongs to those who shift their focus toward strengthening the underlying rails of global systems, which leads to compounding benefits that resist standard hype cycles.
Embedding AI in core business processes
The key to moving beyond isolated pilots is to focus on infrastructure-level applications that solve real, systemic problems rather than chasing generative novelties. In the fintech and payments sector, we have already seen how this works – AI-driven risk scoring, fraud detection and transaction monitoring are not speculative but are instead quietly embedded and continuously improved within the core workflow.
To achieve true scale, organisations must prioritise utility over visibility and ensure AI is integrated into systems that require high reliability and continuity. By building resilient infrastructure and interoperable systems, businesses can ensure that when the initial hype fades, their technological foundations remain intact and functional.
Governance and accountability
Establishing clear governance is essential because the current AI market concentration creates significant systemic fragility.
When valuations run ahead of fundamentals in a narrow part of the stack, the entire system becomes brittle, and any slowdown can reverberate widely. A “Hindenburg-style crash” is a plausible risk when commercial pressure causes organisations to bypass rigorous safety testing.
To ensure resilience, we must have a clear understanding of what we are optimising for before we deploy these tools at scale. Robust governance and accountability ensure that genuine infrastructure-level applications do not get caught in the fallout of overpromising elsewhere in the market.
Tracking long-term business outcomes from AI strategies
We should look to the durable AI models found in financial services to define our metrics, focusing on embedded value that resists hype cycles.
Instead of simple usage rates, we should track outcomes such as reduced fraud losses, improved routing efficiency and the protection of smaller businesses. These systems create compounding benefits across the ecosystem and, crucially, they age well.
Identifying which initiatives survive a market lull or a tightening of capital will expose which tools have true utility versus those adopted only for visibility. Long-term planning compounds, whereas short-term optimisation eventually hits constraints.
Institutional knowledge and scaling AI
It is vital to protect and codify institutional knowledge because the Darwinian process currently at work in the AI sector will eventually separate durable systems from disposable ones. If an organisation relies solely on overconfident generative novelties without grounding them in the long-term thinking required for robust systems, they risk losing their core competitive advantage during a market correction.
Success belongs to those building for longevity, which involves shifting from treating AI as a human-like entity to using it as a tool to strengthen the underlying rails of business systems.
By codifying expertise into shared standards and interoperable systems, businesses ensure that their unique value is protected and that they are building systems designed to last, not just to launch.
