Liontrust’s Uru and Pleydell-Bouverie: When will we see the elusive return on AI investment?

Authors: Clare Pleydell-Bouverie and Storm Uru, Fund Managers, Liontrust Global Innovation Team

If there’s one burning question on investors’ lips this earnings season, it’s when will AI start generating meaningful revenue? The $210 billion in capex planned for 2024 from just the four largest AI infrastructure investors (Alphabet, Meta, Amazon, and Microsoft) justifies that question.

These companies and many others are investing heavily in AI and plan to do even more in 2025. Why? Alphabet CEO Sundar Pichai summed it up perfectly during the company’s second-quarter earnings conference call: “The risk of underinvestment here is much greater than the risk of overinvestment for us.”

That sentence captures two key points that have been echoed in Silicon Valley this earnings season. The first is that the expected gains from AI are worth pursuing, based on evidence from early AI use cases that are already generating compelling returns.

See also: Nvidia stock price falls in over-the-counter trading despite improved forecasts

The second is that investment risk can be managed by generating sufficient cash flow. Microsoft, for example, spent a whopping $44.47 billion on capital expenditures in the last 12 months (up 60% year over year), but that left 63% of its operating cash flow untouched.

So where is that elusive return on investment (ROI)? You probably won’t find it in the financial statements of most companies that tout their AI credentials during earnings conference calls.

Artificial intelligence has certainly become a buzzword, but that doesn’t mean some AI pioneers aren’t seeing tremendous benefits.

Pioneers of artificial intelligence

Meta stands out among the tech giants, and the company’s AI investments in content recommendations and AI advertising tools are already generating tangible revenue. Over 50% of the content we see on Instagram is now recommended by AI, while advertisers using Meta’s AI advantage+ tools see a 22% higher return on ad spend.

These factors helped drive ad revenue up 23% year over year in Q2. This is the ROI everyone is looking for, there are just very few companies that can deploy AI at scale and generate significant revenue.

See also: Choosing ‘Picks and Shovels’: Executives Look Beyond Nvidia for AI Solutions

ServiceNow, a leading enterprise workflow automation software company, is also among that select minority of companies currently implementing AI at scale. After using Now Assist, the company’s Generative AI assistant, internally for 90 days, the company saw more than $10 million in annual enterprise savings.

Gen-AI customers are seeing similarly dramatic productivity gains – BT Group reduced the time it takes agents to write and review tickets by 55%. Now Assist has become the fastest-growing product in the company’s history, and customers are paying.

Beyond Big Tech

For smaller companies, significant AI-related revenue is harder to pin down but can be found. These include Palantir, whose AI-powered AIP platform improves data analysis and operational efficiency for corporations and defense organizations. Using AIP, General Mills achieved $14 million in annual cost savings. Demand for Palantir’s AI solutions is not only driving revenue growth (up 30% year over year in Q2), but it has also turned the company profitable over the past seven quarters.

One thing all of these companies have in common is world-class compute infrastructure. For example, Meta trained its latest large-language model, Llama 3, on two GPU clusters, each consisting of 24,000 Nvidia H100s. Llama 4 will be trained on 10x more compute power. This is the infrastructure needed to deliver the next step in model performance, which will enable further revenue opportunities.

Not only do we believe this investment is justified, but it is also a rational allocation of capital. For every $1 invested by cloud providers in Nvidia-accelerated compute infrastructure (GPUs and the associated stack required to build and train AI), it will translate to an estimated $5 over four years.

See also: Tech giants face an ‘existential event’

This is because AI solutions are moving customers to the cloud, which is driving revenue growth for vendors, which will translate into infrastructure ROI. For companies that support AI models (basically provide an AI inference kit), the ROI is even more significant—$1 invested today has the potential to generate $7 in revenue over a four-year time horizon. It’s this compelling AI infrastructure ROI that’s driving companies to accelerate AI infrastructure investments despite recent investor turmoil.

AI-generated revenues won’t be available to everyone. AI is a platform technology shift that, like all replacement platform shifts, will bring both disruption and opportunity. And companies need to build that infrastructure before they can monetize it.

As such, we believe that the value for investors currently lies primarily in the AI ​​infrastructure layer of the new technology stack. Over time, the baton will be passed to the AI ​​application layer.

We expect this baton to be passed more quickly than with previous technology shifts, because the payoffs have the potential to be much greater. It took Microsoft seven years to reach $3.5 billion in annual revenue thanks to its move to the cloud; the company reached that milestone in just 18 months thanks to generative AI.