That distinction may sound subtle. It is not. It is the difference between organizations that get surprised by deep tech and those that don’t. And it is the single biggest lesson I have drawn from our quantum journey — a lesson that, I am now convinced, applies to deep tech as a whole.

Most of us in the technology industry built our instincts in the digital era: fast cycles, rapid iteration, first-mover advantage. Deep tech follows a different physics. It is science-rooted, slow-maturing, and profoundly non-linear. Progress looks discontinuous in the headlines — a “breakthrough” one month, a “winter” the next — while underneath, the real work advances steadily, cumulatively, mostly through unglamorous engineering.

This is why the digital playbook fails when applied to quantum, and to deep tech more broadly. Success here depends far less on speed, and far more on readiness, timing, and positioning.

For end users of deep tech, the hard question is not “what’s the use case?” or “where’s the ROI?” — those questions come later, and forcing them early is one of the most common mistakes I see. The hard question is: when do you step in?

Step in too late, and you risk being caught off guard, ceding strategic ground you may never recover. Step in too early, and you risk premature commercialization: over-investment, loss of momentum, and damaged credibility when promised returns fail to materialize.

There is no single right moment. There is only informed positioning, aligned with how mature the technology actually is — and with how much you stand to win or lose. What that positioning looks like in practice: building readiness for technologies that may take years to mature, developing enough in-house prowess to separate signal from noise, engaging selectively with research, and weaving ecosystems with partners, start-ups, and academia. The early value is not revenue. It is better decisions and reduced uncertainty. That value is real, and it compounds.

Over years of quantum work, we have seen deep tech move through three distinct phases — and seen how badly things go when organizations confuse them.

• The technology phase is about maturing the technology itself. Expect a steady stream of “breakthrough” announcements; treat them as data points, not turning points. The sensible moves at this stage are horizon scanning, feasibility analysis, understanding constraints and risk drivers, and controlled experimentation. The goal is to be ready for the next phase — not to force it.
• The application phase begins as credible application directions start to surface. This still does not mean full deployment. It means integrating into early workflows and testing assumptions in controlled environments. One thing experience has taught me: the strongest applications are rarely the obvious ones. Surprises are common — and often where the value hides.
• The business phase only becomes central once applications stabilize. The focus then shifts to repeatability, economics, trust, and integration. The most common failure mode in deep tech is jumping to this phase before feasibility is established. We have all seen the casualties.

Quantum is still largely in the technology phase — but not uniformly so, and this nuance matters. In AI simulation-led domains, we have already entered the application phase, and value is being created today. In drug discovery, we are working with GSK for example, on how quantum simulations can generate the high-quality data that AI models need to learn faster — because AI is a data-hungry beast, and producing that data experimentally is slow and expensive. In materials science and chemistry, quantum-mechanics-informed simulation is already feeding AI-driven R&D workflows for batteries, alloys, and chemical processes, delivering insights that black-box AI cannot produce on its own. These are not demos. They are early workflows creating real scientific and business value — exactly what the application phase looks like.

Elsewhere, the sensible work remains readiness work — and here too, the signals are getting clearer.

Take a project recently announced with NESO, the National Energy System Operator for Great Britain, led by our colleagues at Cambridge Consultants. The team is performing a deep technology assessment of quantum hardware to quantify the risk quantum computing poses to the cryptography protecting critical energy infrastructure. Note what this is and isn’t: it is not about deploying quantum computers. It is about preparedness in a regulated, high-stakes environment — exactly the kind of work that fits this moment.

The discipline underpinning this work is resource estimation, which we will be exploring in our upcoming blog, From Quantum Exploration to Execution. Resource estimation turns ambition into technical reality: qubits, error rates, timelines, and scaling behavior become explicit rather than aspirational. It allows organizations to distinguish feasible paths from dead ends, prioritize actions in proportion to actual risk, and avoid the twin traps of hype-driven paralysis and premature bets. We apply the same approach with Airbus on quantum for materials science, and with governments worldwide on readiness and strategic risk. Resource estimation is not an academic exercise. It is a strategic readiness capability.

If this pattern sounds familiar, it should. Large language models were developed over many years of steady, mostly invisible progress. When generative AI hit mainstream adoption, many organizations were blindsided. Others were not — because they had engaged during the technology phase, built internal AI literacy and intuition, and experimented without forcing early commercial outcomes. When the inflection point arrived, they adapted workflows faster, integrated with less friction, and moved from experimentation to impact while others were still forming task forces.

That is the lesson in one line: early engagement does not mean early commercialization. It means being ready when the inflection point arrives.

At Capgemini, we think of deep tech as a continuous innovation spectrum: sense and explore, experiment and de-risk, pilot and prepare, industrialize and scale. The right actions differ at every point along that continuum, depending on technology readiness, risk exposure, and what is genuinely at stake. Success comes from operating across the whole spectrum — not from optimizing a single stage. Quantum is simply the clearest example today of this thinking in practice.

Deep tech is a long game. After four years of living it through our Quantum Lab, I am more convinced than ever: the winners will not be the fastest movers, but the best prepared. The playbook — not the headlines — determines the outcome.