The companies winning with AI are not the ones cutting costs

Every business I have worked with on AI this year has come in thinking about growth. New revenue, new capabilities, things they could not do before. I have started to realise this is an unusual starting point.

The AI story in the news is redundancies. 150,000 tech jobs gone so far this year, with AI cited as a reason in nearly half of those cuts. Your board is asking about headcount. Your CFO is asking about cost per output. Every ambient signal suggests that AI is a shrinking exercise - a way to do what you already do with fewer people and lower costs.

This is not wrong. But it is incomplete. And the gap between those two things is where competitive advantage is being built right now.

The finding buried in the productivity data

PwC’s 2026 AI Performance Study interviewed 1,217 senior executives across 25 sectors and found something that the efficiency headlines have largely buried: 74% of all AI economic value is being captured by just 20% of companies.

The question worth asking is: what do those companies have in common?

It is not the size of their AI budget. It is not which tools they have deployed or how many pilots they have run. It is the question they started with.

The companies in the top 20% asked: what can we do now that we could not do before? Most of the rest are asking: how do we do what we already do more cheaply?

Both questions lead to real results. Only one leads to compounding advantage.

The IKEA example worth understanding properly

IKEA’s Billie chatbot is one of the most cited AI implementations in retail right now. It handled a volume of customer service enquiries that would previously have required thousands of agents, and delivered a clear financial result: €13 million in savings.

That is a real outcome. It is also the less interesting half of the story.

Instead of cutting the 8,500 people the technology had freed up, IKEA asked a different question: what can these people do now that they could not do before? The answer was remote interior design consultation - a service that could not have been delivered at scale without the cost savings AI had just created.

The same programme that saved €13 million generated €1.3 billion in new revenue.

Both outcomes came from the same AI implementation. The difference was entirely in how the question was framed at the start.

Why most boards land on the wrong question

The efficiency frame is not irrational. It is what the evidence in front of most boards supports.

Redundancy headlines are visible and concrete. The business case for cutting cost with AI writes itself - you can model it, track it, and report it in a straightforward line on the P&L. The growth opportunity is more abstract. It requires a different kind of strategic thinking, a willingness to redirect capacity before you know exactly where it will go, and a longer measurement horizon.

There is also a timing problem. Cost savings from AI are immediate. Revenue opportunities from AI-enabled capabilities take longer to build and are harder to attribute. In a quarterly reporting environment, the efficiency play will always look more compelling in the short term.

This is exactly why the gap between the top 20% and the rest is widening. The companies pointing AI at growth are making investments that do not show up cleanly in this quarter’s numbers. They are building capabilities their competitors will be trying to reverse-engineer in two or three years. By the time the gap is visible, it is very hard to close.

It is also worth noting that AI amplifies what it touches - which means pointing it at an efficiency-first question tends to entrench that direction of travel. The strategic framing you start with shapes what your AI programme becomes.

What do the companies winning with AI have in common?

Looking at the businesses in the top 20%, the pattern is consistent. PwC found that AI leaders are two to three times more likely than peers to use AI to identify and pursue growth opportunities - and twice as likely to redesign workflows around AI rather than simply adding tools to existing processes.

Four things stand out:

1. They start from the customer, not the process. The efficiency question begins with an internal process and asks how AI can reduce the cost of running it. The growth question begins with the customer and asks what they need that you currently cannot provide. IKEA did not start with “how do we reduce call centre headcount.” They started with “what do our customers actually want help with.”

2. They treat freed capacity as an asset, not a saving. When AI reduces the time your team spends on a task, the default financial response is to reduce headcount. The growth response is to ask what that freed capacity can be redirected toward. This is not a soft argument about looking after your people - it is a hard calculation about where the value goes next.

3. They think in capabilities, not tools. The efficiency mindset drives tool procurement: what AI can we buy to reduce cost X? The growth mindset drives capability building: what can we now offer or do that was previously impossible or unaffordable? These feel like the same conversation but they lead to completely different investment decisions.

4. Their AI roadmap connects to revenue, not just headcount. Not every line item needs to link directly to new revenue. But if you read your AI roadmap and the primary success metric is cost reduction or headcount reduction, the direction of travel is already set.

The hiring data is also worth your attention

Alongside the economic value finding, PwC’s separate 2026 AI Jobs Barometer - which analysed more than one billion job advertisements across 27 countries - contains a number that is almost entirely underreported: junior roles requiring AI skills grew 35% since 2019, while other entry-level roles shrank 10%.

What that tells you is that the companies investing in AI capability - not just AI tooling - are creating roles, not just eliminating them. The skills mix is shifting. But the headcount direction depends entirely on what you are pointing AI toward.

If you are building toward efficiency, AI functions as a replacement technology. If you are building toward growth, AI functions as an enabling technology. The same models, the same tools, completely different trajectory for your business two years from now.

A practical audit worth doing

Take your AI roadmap - or if you do not have a formal roadmap, the last three AI tools or projects your business has invested in. For each one, answer one question:

Does this open something we could not do before, or does it make something we already do cheaper or faster?

There is nothing wrong with the second category. The question is the ratio.

Most leadership teams, when they do this exercise, find the balance more lopsided toward efficiency than they expected. That is useful information, and the starting point for a different conversation about where the next round of investment should go.

If you want a more structured way to work through this, the CEO’s guide to AI implementation covers the sequencing that actually produces returns - starting from the right question, not the easiest one.

Final thought

The redundancy headlines are not wrong. Plenty of businesses are using AI to cut costs and it is working. What the data shows is that this is the floor, not the ceiling.

The 20% of companies capturing 74% of AI economic value are not doing something exotic. They are asking a better question at the start. What can we do now that we could not do before?

Productivity is a real outcome. It is also, increasingly, the outcome that every business with access to an AI tool can achieve. If you are competing on productivity alone, you are competing on a metric that is fast becoming a commodity.

The businesses that will look back at 2026 as the year they pulled ahead are the ones that treated freed capacity as the starting point, not the destination.

Frequently asked questions

Why are some companies getting much more value from AI than others?
PwC’s 2026 AI Performance Study found that the top 20% of companies capture 74% of AI’s economic value - not because they have bigger budgets or better tools, but because they point AI at growth rather than just efficiency. They are two to three times more likely to use AI to identify new revenue opportunities and twice as likely to redesign workflows around AI rather than layering it onto existing processes.

Is AI better for cutting costs or driving growth?
Both are achievable, but the data shows a significant difference in outcomes. Companies focused primarily on cost reduction get real savings - but those focused on growth with AI generate returns that compound over time. The IKEA Billie case is instructive: the same AI implementation saved €13 million in operational costs and generated €1.3 billion in new revenue, depending entirely on what question the business asked about the freed capacity.

How do I know if my AI strategy is focused on the right things?
Take your last three AI investments and ask one question about each: does this open something we could not do before, or does it make something we already do cheaper or faster? Most leadership teams find their portfolio more lopsided toward efficiency than they expected. That ratio is the starting point for a better conversation about where the next round of investment should go.

What does an AI growth strategy actually look like in practice?
The companies in PwC’s top 20% share a few consistent traits: they start from the customer rather than the internal process, they treat AI-freed capacity as something to redirect rather than cut, and they build toward new capabilities rather than cheaper versions of existing ones. The CEO’s guide to AI implementation covers the practical sequencing in more detail.

Connected Paths works with CEOs on AI strategy and implementation - starting from the right question, not the most obvious one. Start here.