Proximity isn't performance: turning AI investment into TMT value

Yet, even here, the gap between AI investment and AI value persists.

“The challenge for TMT leaders is rarely a lack of spending or ambition; it’s the ability to tie that investment back to execution and measurable business outcomes.”
Jakob Haesler Consulting Leader, Forvis Mazars Group

That gap matters more in 2026 than ever before. As our AI at scale: from experimentation to real business value report explores, boardrooms have moved from asking what AI is being explored to interrogating what value it has actually delivered. For a sector that positions itself at the forefront of the AI boom, the pressure to demonstrate returns is correspondingly higher.

The good news is that some TMT organisations are doing exactly that, and the patterns behind their success are increasingly clear. While each sub-sector is leveraging AI and differently, the most successful, AI-value-generative organisations are using AI as a tool to solve a problem, not as an aim in and of itself, and implementing well-foundationed AI applications as a result that create real business outcomes.

Three sub-sectors, three very different AI stories

AI adoption is moving quickly across TMT, but not uniformly. Each sub-sector within – technology, media and telecommunications – is leveraging AI in distinct ways, with distinct levels of maturity.

Technology

This is the sub-sector even laypeople think of when considering AI: the hardware and software providers that power every aspect of an increasingly digital world.

These organisations are leading the conversation around AI value, and those seeing the best ROI are implementing on two fronts: embedding AI into products and into operations simultaneously. Software companies increasingly build AI directly into the product itself rather than offering it as an adjacent feature, and the most effective implementations are those where the software's data is meaningfully integrated into the AI functionality. This could be in user-facing applications like in-app analysis, or in back-office functions like automated load underwriting. In hardware, meanwhile, AI has become the driver of innovation itself, a newer dynamic that is reshaping product roadmaps across the sub-sector.

The disruption cuts both ways, however. The combination of agentic AI and large language models in software development has fundamentally changed how SaaS businesses operate.

“Where engineering teams once spent the vast majority of their time coding and a fraction on data, that ratio has effectively inverted. The focus is now on data engineering and making data work reliably, and the traditional model of monthly subscriptions funding periodic updates is under genuine threat.”
Guillaume Devaux Partner, Head of Technology, Media, Telecommunications, Forvis Mazars Group

Underneath all of this sits a strategic race: for chips, for market position and for control of the sector's direction. Questions of data sovereignty and autonomy are becoming increasingly pressing, with the most resilient supply chains being those where each link can operate independently even while remaining deeply connected.

Media

Media and entertainment is using AI extensively for content production and personalisation, with advertising at scale a particular focus. Indeed, some of the earliest embedded AI successes came from this space – such as personalised recommendation engines powered by machine learning – long before the current wave. But maturity remains very inconsistent across the sub-sector, and copyright is an ever-present concern that shapes what organisations can responsibly deploy.

Telecoms

Telecommunications is arguably the quiet achiever. Telecoms operators are fairly mature in their operational use of AI, applying predictive models to traffic management and outage monitoring and deploying AI-driven customer service at scale. Their business models naturally lend themselves to extracting value from AI, and partnerships have proven hugely important in accelerating that journey.

What value generators are doing differently

Across all three sub-sectors, the organisations seeing genuine returns share a common starting point: they use AI tooling to solve a problem they already wanted to solve. AI is a powerful tool for addressing strategic problems, not an end in itself, and implementations that follow the technology rather than the need rarely deliver genuine business value.

Data readiness is the second major differentiator. AI-native organisations enjoy a consistent data model from day one, but those coming from legacy environments find their data scattered and inconsistent, requiring real investment before the technology can be reliably enabled. The more decentralised the organisation, the more acute this challenge becomes. Larger organisations that are investing heavily in truly embedded AI (rather than bolting on off-to-the-side chatbots, or other disconnected use cases) are the ones converting capability into performance.

A structural shift is underway in how technology (and especially software) is leveraged: AI is making the conversation less about platforms and more about applications. The single biggest impact area identified across the sector is software development itself, where AI can massively simplify what has historically been an enormous organisational burden. With the cost and effort of building software falling so dramatically, the economics that once favoured sprawling, full-service solutions are weakening. Why settle for a one-size-fits-all platform, much of which goes unused, when a focused application can be built and tailored to solve a specific business need?

The value generators are leaning into this shift, developing customised applications shaped around individual problems, customers and workflows. A common mistake is the inverse: attempting to apply the same AI feature uniformly to every customer or function, when the real opportunity lies in precision.

Stewardship as strategy

For TMT organisations, being a good steward of AI is not simply an ethical obligation; it is fast becoming a commercial one.

The regulatory picture – spanning GDPR, the EU’s AI Act and beyond – is still evolving, and uncertainty remains about how enforcement will play out in practice. Discrepancies between US and EU approaches create genuine apprehension among even the largest players – and especially these players, as multiple jurisdictions are likely to apply.

But as frameworks solidify, compliance is emerging as a competitive advantage.

“Companies that can justify their AI-powered decision-making, be transparent about training data and usage and demonstrate a risk-based approach to AI implementation will be better positioned than those treating compliance as a final hurdle.”
Will Stallings Director, Forvis Mazars US

That risk-based approach should extend to operational concerns, too. Placing AI in customers' hands introduces new exposure, not least in cybersecurity, and the vast quantities of data these products collect carry responsibilities of their own. Not to mention the ecological implications and impacts of AI, which are vast. Encouragingly, frontrunners in the sector are already embedding mindful principles from the outset, with ethics-by-design groups working to ensure responsible use is built in rather than retrofitted.

Who will win the real AI race?

TMT organisations have a structural advantage in the AI era: proximity to the technology, data-rich business models and customers who expect innovation. But proximity is not the same as value. The organisations converting AI investment into measurable returns are those treating it as a strategic discipline, grounded in real business problems, integrated data and compliance-by-design, rather than a race to deploy.

In TMT, the race that matters is not who implements AI the fastest. It is who implements most effectively in a way that generates real value, both for themselves and for their customers.