The AI trade has been one of the most dominant stories in markets over the past two years, but voices calling for caution have grown louder as valuations stretch and the technology’s real-world returns remain uneven. Against that backdrop, UBS Global Wealth Management’s Nadia Lovell is doubling down, laying out what she calls the “sweet spot” for AI investment.

Lovell, speaking on behalf of the firm, remains bullish on artificial intelligence. She points to three specific areas that she believes will drive returns: platforms, applications, and pricing power. The thesis is concise but carries weight coming from a global wealth manager responsible for advising high-net-worth clients on where to put their money next.

Platforms: the infrastructure bet

When Lovell talks about platforms, she is referring to the foundational layers of AI — the cloud computing services, the large language model providers, and the hardware ecosystem that makes AI possible. These are the companies building the pipelines and the compute capacity. Chipmakers, hyperscale cloud providers, and model developers fall into this bucket. The reasoning is straightforward: no matter which application wins, the platforms underneath will collect rent. AI models need training and inference, and that requires massive data center infrastructure.

This is not a new idea, but it remains the most tangible way to invest in AI right now. Platforms generate revenue today. They have customer contracts, usage-based pricing, and clear growth trajectories. Companies in this space have already reported surging capital expenditures and equally surging revenues tied to AI workloads. The risk is that competition could compress margins, but so far the largest players benefit from scale that newcomers struggle to match.

Applications: where the value crystallizes

The second pillar of Lovell’s sweet spot is applications. These are the software products and services built on top of AI platforms. Think of enterprise tools, productivity suites, code generators, customer service chatbots, medical imaging analysis software, and the long tail of industry-specific AI tools. Applications are where AI moves from a raw capability to a solution that solves actual problems for businesses and consumers.

Lovell’s focus on applications suggests she sees the next wave of AI value creation coming from companies that can embed AI into workflows and deliver measurable productivity gains. Investors have been waiting for a killer app that justifies the infrastructure spend, and a handful of candidates have emerged — from writing assistants to coding copilots to drug discovery engines. The challenge is that the application market is fragmented, with many startups competing alongside established software incumbents. Margins can be thin until a product gains network effects or becomes stickier than rivals.

Still, the logic holds: if platforms are the picks-and-shovels play, applications are where the gold gets refined. Lovell’s inclusion of apps in the sweet spot signals that she believes the market is past the pure infrastructure hype phase and entering a period where specific use cases will separate winners from losers.

Pricing power: the financial reality check

The third factor — pricing power — is perhaps the most revealing. Lovell is not just looking at revenue growth; she is looking for companies that can charge more for their products over time without losing customers. In AI, this is a difficult trick. Many AI tools are initially offered at low prices to build adoption, and competition has pushed some vendors into price wars. But the winners will be those whose AI solutions are so valuable that customers are willing to pay a premium — and keep paying.

Pricing power signals a strong competitive advantage. It could come from proprietary data, unique algorithms, or a deeply integrated product that becomes part of a customer’s daily operations. Switching costs matter. When a company’s AI tool is woven into a legal firm’s document review process or a hospital’s diagnostic pipeline, the price becomes a secondary concern compared to the cost of switching.

Lovell’s mention of pricing power suggests she is filtering out companies that would be commoditized by open-source alternatives or by rivals with deeper pockets. She wants the ones that can raise prices year over year, a trait that historically correlates with long-term stock outperformance.

Putting it together

These three elements — platforms, apps, and pricing power — form a screening framework. Not every AI company will check all three boxes, and Lovell’s sweet spot is likely the intersection. A cloud provider owns the platform, has a suite of enterprise apps, and can raise prices because customers are locked into its ecosystem. A niche AI startup might have a great app but no platform of its own, and pricing power might only last until a larger rival copies the feature. An infrastructure company might have platform dominance but no direct app presence, leaving it dependent on the success of others.

The implication is that the highest-conviction AI investments will be hybrid players or ecosystem owners. This aligns with the current market leaders — the trillion-dollar names that have built both the chips, the cloud, and the application layer. But Lovell’s framework also leaves room for smaller companies that can carve out a defensible niche in one of the three pillars.

Counterpoints and risks

No bull case is complete without acknowledging the other side. AI stocks have run hard. Valuation multiples are elevated relative to history, and any disappointment in earnings or AI adoption could trigger sharp corrections. Regulatory risks are also mounting. Governments in the US, the EU, and China are drafting rules around AI safety, copyright, and data privacy. Tighter regulation could slow deployment and increase compliance costs, particularly for application-layer companies.

There is also the question of timing. Platforms are already priced for perfection. A slowdown in enterprise AI spending — which could happen if the broader economy weakens — would hit infrastructure companies first. Applications, by contrast, may take years to reach the scale that justifies today’s valuations. Pricing power, too, is fragile. In a recession, corporate customers become more price-sensitive, and the luxury of raising prices evaporates.

Lovell’s response to these risks, as implied by her framework, is that focusing on all three factors provides a margin of safety. A company with platform dominance, a growing app portfolio, and pricing power is better equipped to weather downturns and regulatory changes than a single-product vendor with no moat.

What it means for investors

For individual investors, Lovell’s sweet spot offers a concrete way to evaluate AI plays beyond the hype. Instead of buying every stock that mentions AI in its earnings call, ask: is this a platform, an app, or neither? Does it have pricing power? If the answer to two or three of those questions is yes, it might be in the sweet spot.

That does not mean ignoring valuation. Even the best AI company can be a bad investment if bought at the wrong price. But Lovell’s framework shifts the conversation from “is AI real?” to “which AI companies have lasting advantages?” It is a mature, selective approach at a time when the market needs one.

UBS Global Wealth Management publishes regular guidance for its clients, and Nadia Lovell’s remarks are part of that ongoing advisory. For now, the firm sees the AI trade as having legs — but only for those who know where to look.