As companies finalise their 2026 strategies, we are witnessing the emergence of a defining split in how businesses approach artificial intelligence. On one side are firms integrating AI into their growth plans with strategic rigour, on the other are those treating it primarily as a cost-cutting exercise.
Next year will expose just how wide that divide has become.
The stakes are higher than many business leaders realise. McKinsey reports that roughly 70 per cent of AI pilots fail and the pattern is consistent as these failures occur when AI projects are not tied to business goals, mission or vision. I often describe using a Formula One analogy. AI is the car – powerful, sophisticated and capable of extraordinary performance. But without the racetrack – the strategy – it is essentially useless.
Successful AI adoption works from the top down. Mission and vision cascade into strategy, strategy informs workstreams and workstreams generate the enabling AI use cases. Without that alignment, you are deploying technology in search of a problem rather than building solutions that deliver measurable business outcomes.
A market shake-out is coming
The AI consulting market is heading for consolidation in 2026. Smaller startups and consultancies will struggle as hyperscalers and major players begin to enter the market at scale and dominate. However, we are also seeing a structural change in how companies procure AI services. Rather than the £50 million mega-deals that made headlines previously, I expect to see numerous transformation projects in the £500,000 range as companies break their AI journeys into manageable phases.
AI transformation is not a single project. It is the alignment of people, processes and technology across multiple moving parts. Companies are learning that incremental progress with clear milestones beats ambitious programmes that collapse under their own complexity.
Hyperscalers and investment firms are already keeping a watchful eye on data centre strategy because of AI’s impact. That is becoming a critical trend as the environmental cost of AI computing is substantial, forcing a rethink of infrastructure at scale.
Growth, not just savings
Goldman Sachs’s CEO recently stated he would double his budget and spend billions on AI, not to cut headcount, but to drive growth. That signals a split between companies using AI strategically for expansion and those simply chasing cost savings.
The firms focused on growth understand that AI is a catalyst for better processes, faster decision-making and enhanced productivity. Their focus is on enabling people to work more effectively, rather than replacing them. After a challenging economic cycle, AI offers a clear path to foundational change that unlocks new revenue rather than simply trimming expenses.
Those who reach at least the first phase of transformation should see noticeable growth by 2026. But those using AI only for simple cost reductions will find themselves left behind, having invested in technology without building the capabilities needed to extract real value.
Regulation will tighten
Failures, such as the recent Deloitte Australia case, will likely accelerate regulatory scrutiny. Deloitte was required to refund part of a $440,000 fee after delivering a government report that used AI-fabricated content that was missed by governance processes.
If such issues are not systematically checked, more regulation is inevitable. The challenge is that regulatory approaches vary considerably. The UK currently favours principles-based regulation, whilst the EU is more prescriptive. International compliance is complicated and companies operating across borders face differing requirements by country and sector.
As such, I expect increasing regulatory scrutiny over the next year. Companies need robust checks and balances, not just to comply with emerging rules, but to avoid the reputational damage that comes from deploying AI without proper oversight.
Smaller models, bigger impact
I don’t envision a future where massive AI behemoths consume enormous compute resources. 2026 will see movement towards smaller, domain-specific models that reduce both costs and environmental impact.
These models are more efficient for specific tasks and help address the sustainability concerns that are beginning to dominate data centre strategy. For many firms, this means AI becomes increasingly accessible and more practical to deploy at scale without the infrastructure burden of general-purpose large language models.
We can also expect to see much more experimentation next year. AI is beginning to move beyond spotting patterns and towards generating ‘novel insights’, which are genuinely more insightful outputs. It is still early, but I think we will get a lot closer over the next year.
A big conversation in 2026 will be about the capabilities of smart AI. We are not quite there yet, but we are moving in that direction. I don’t think superintelligent or sentient AI is something for next year. Still, the transition from pattern recognition to more autonomous, insightful systems is real.
For any of this progress to happen, companies need to get the fundamentals in order – their data structure, infrastructure, cloud setup and, most importantly, their governance.
As we enter 2026, the question isn’t whether firms have AI capability, it’s whether they’ve built the strategic foundations to actually use it. The gap between those who get this right and those who don’t will widen considerably this year. That divide is already forming and there’s no neutral ground. You’re either building the racetrack or polishing a car that’s going nowhere.
