Google Cloud: Why AI Strategy Beats Technology Alone

Enterprise AI adoption has reached 78% globally, yet a troubling gap has emerged between implementation and value creation.

More than 80% of companies report no measurable impact on their bottom line from AI investments, while 42% of executives acknowledge that deployment efforts are creating internal disruption rather than business benefits.

The performance divide is stark. Companies with comprehensive AI strategies achieve 80% implementation success rates, compared to just 37% for organisations taking piecemeal approaches.

2025: The year chatbots grew up

Oliver Parker runs global AI sales for Google Cloud, which means he spends his days talking to executives who are either thrilled with their AI results or bewildered by them.

He’s watched as the technology evolved from research previews of chatbots into something far more capable.

“These AI systems started very much as chatbots over the last couple of years,” he explains. “The agent-based capabilities are really around automating lots of new things versus question and then answer, which is how LLMs started.”

The difference matters more than it sounds. Early AI tools could answer questions or generate text, but they couldn’t actually do anything useful with that information.

Now they can take action, remember context from previous interactions and handle complex tasks that used to require human intervention.

In banking, for example, “the agent can understand who you are and actually talk about your spending and your savings plan, right the way through the system,” Oliver explains.

Today, Gen AI is driving serious money. The cloud AI market hit US$78.36bn last year and analysts expect it to reach US$589.22bn by 2032.

But most of that growth is coming from a relatively small number of companies that have figured out how to make AI actually work. The rest are burning cash on projects that look impressive in presentations but don't move the needle.

Why AI strategy actually matters

The performance gap between companies with proper AI strategies and those without is staggering. Businesses that have taken time to develop comprehensive AI plans report 80% success rates with their implementations. Those who haven’t, just 37%.

The failures aren’t about picking the wrong technology or hiring the wrong consultants. Research shows that 68% of executives report serious friction between their IT departments and business teams when it comes to AI projects.

Even more telling, 72% say their AI applications are being developed in isolation, with different parts of the company building competing solutions that don't talk to each other.

The money follows the same pattern. Companies that invest properly in AI see success rates 40 percentage points higher than businesses that don’t. Half-hearted attempts at transformation are worse than doing nothing at all, because they consume resources whilst delivering disappointment.

Skills shortages add an extra challenge. A third of enterprises report they simply don't have people who understand how to implement AI effectively, while more than half are paralysed by data privacy concerns.

Google’s bet on intelligence

Google Cloud sits in an interesting position in this market. With between 10% and 13% market share, it's smaller than Amazon Web Services (AWS) or Microsoft Azure, but it's betting that its AI background gives it an edge. 

“We’ve been an AI-first company for a very long period of time,” Oliver notes. “When you think about our mission statement in terms of organising the world’s information and making it universally accessible and useful, that is fundamentally a data and AI mission statement.”

The company is trying to compete across the entire stack, from computing infrastructure through to ready-made AI applications.

“We have infrastructure for people that want to build on our AI infrastructure,” Oliver explains.

“We then have models: our Gemini models, other models like Lyria and Chirp that are very specific to certain tasks. We have a platform so developers can build on that platform but take advantage of not just our models, but anybody else’s models.”

Google’s newest product, Agentspace, takes a different approach to enterprise AI adoption. Instead of requiring companies to rework their entire technology stack, it tries to embed AI capabilities into the tools people already use.

It’s a recognition that most successful AI implementations happen when they solve immediate problems for real users, rather than requiring wholesale organisational change.

The companies getting it right

The best way to understand what works is to look at companies that have moved beyond pilots to genuine business impact. Oliver points to Screwfix, the British trade tools retailer, which built an AI-powered visual search system for its website.

“I sat down with them nine months ago at the London Summit and they were just trialling it. Now, 50% of the users on the B&Q website are using Screwfix Lens.”

Similar patterns are showing up across industries. Three-quarters of manufacturers have deployed AI, mostly for production optimisation, customer service and inventory management.

Nearly 90% of game developers use AI agents. But the successful implementations share common characteristics: they solve real problems, they’re easy to use and they deliver measurable value quickly.

The economics have shifted dramatically in favour of these practical applications. The cost of running AI models has dropped dramatically: between November 2022 and October 2024 the cost to run AI at GPT-3.5’s level of performance dropped by more than 280x.

That economic transformation has removed the cost barriers that used to limit AI to experimental projects, opening up routine business applications that were previously uneconomical.

What comes next?

Oliver sees two major trends shaping enterprise AI adoption. First, AI capabilities are moving from specialised applications into everyday work tools.

“You’ll start to see those kinds of capabilities show up in many more companies for their employees,” he says. “Agentspace is really designed to put the capabilities that are at product level into the hands of employees.”

Second, customer-facing AI applications are becoming more sophisticated and widespread.

“With the capability of these models and the accuracy that’s starting to come from these models, you’re going to start to see lots more external interactions with systems that are AI-based,” Oliver notes.

The companies that succeed in this environment share a few characteristics. They treat AI implementation as organisational transformation, not technology procurement.

They invest in training and change management alongside software licences. They start with specific business problems rather than generic AI capabilities. And they have senior leadership that understands the difference between buying AI tools and actually using them effectively.

Most importantly, they recognise that the window for competitive advantage is narrowing.

As Oliver puts it: “AI represents such a tectonic shift for employees and businesses to accelerate the impact of what they do. Whether you’re an employee wanting to be more productive, or whether you want to change the business that you’re in: there’s no such thing as being too late, but you’ve got to start now.”