European Equities
The AI roadmap: Finding opportunities from evolving innovations
7 min read 30 May 25
Follow us
After enjoying a powerful rally in recent years, AI and technology stocks have had a volatile start to 2025. Concerns about the impact of tariffs, growing competition, and the uncertain pace of AI adoption have evolved to be key issues for this critical sector. Following the latest quarterly earnings season, fund managers Jeffrey Lin and Thomas Lee reveal some of the key developments in the AI landscape and explain why they are optimistic about the potential growth and innovation in the sector.
The most recent quarterly earnings reports from US AI and technology firms provided some interesting insights about trends in the sector.
Product innovation remains strong
The first important message from the recent earnings season, in our view, is that AI innovation continues. Notably, generative AI is moving towards agentic AI, a type of AI that is designed to operate independently. Agents can set goals and take actions autonomously, learning and improving their performance over time.
This trend was demonstrated most prominently by software company ServiceNow, which is rapidly adding agentic AI functions offering clear return on investment (ROI) for its customers.
On the earnings call, ServiceNow executives explained that their new AI-enabled core business suite quickly transforms core business processes, such as HR, procurement, finance and legal. They noted that it's already handling over 135 million employee requests per year. In addition, the company is working closely with customers to help them innovate in their businesses and adopt AI.
Solid corporate results
A number of leading AI firms such as Axon, Meta Platforms, Microsoft, SAP and ServiceNow reported solid earnings results. These companies are seeing AI driving topline growth and margin expansion.
Microsoft noted that it continued to see strong demand for cloud and AI offerings, which it described as “essential inputs for every business to expand output, reduce costs, and accelerate growth”. In the first three months of the year, the revenues for its cloud business Azure grew 33%, year on year, with 16 points of that growth coming from AI services1.
“It seems as if Meta’s investment in AI is really delivering.”
In Meta’s results call, CEO Mark Zuckerberg talked about how AI could transform advertising by improving recommendations. He highlighted a new ads recommendation model, which increased conversion rates by 5% in the quarter. Zuckerberg also said that better content recommendations were leading to users spending more time on Meta’s social media platforms2. It seems as if Meta’s investment in AI is really delivering.
AI spending remains robust
Linked to the growing use and potential benefits of AI, the commentary from management teams at US mega-cap firms around AI capital spending (capex) was positive. It went some way to dispel the notion that demand for AI compute is slowing.
In our view, concerns about potential reduced demand are overblown. The idea stems in part from the release of the DeepSeek AI model in January. The Chinese firm’s R1 model is reportedly capable of achieving similar outcomes as the leading Western AI models from the likes of OpenAI and Google, despite using far fewer chips and being cheaper to run.
We think it’s important to note that most of the “compute efficiency” comments this year have been coming from China. Chinese companies have been forced to do more with less for quite some time due to US export restrictions on cutting-edge technology, particularly from Nvidia.
“Ongoing investment in AI reinforces the idea that AI is going to keep increasing as the technology develops.”
Significantly, none of the large US companies have mentioned a slowdown in compute spending. In fact, in this earnings season, US firms such as Amazon, Alphabet (parent company of Google), Meta and Microsoft have all said that AI capex spending will remain a key priority for them.
For example, Meta raised its capex guidance for 2025 by almost US$6 billion, due partly to additional investments in data centres to support its generative AI efforts. The company expects the investment in infrastructure will give them an advantage in the quality and scale of AI services it can deliver3.
Meanwhile, Google stated that, irrespective of the macroeconomic outlook, it is committed to maintaining its capex plans ($75 billion) for this year. The company said it sees a tremendous opportunity across the organisation and is investing in long term and innovation to drive productivity and efficiency4.
The AI roadmap
In our view, this ongoing investment in AI reinforces the idea that AI is going to keep increasing as the technology develops. In recent months, tech executives, most notably Jensen Huang, CEO of Nvidia, have presented compelling reasons why compute will actually increase in the future.
At GTC 2025, the company’s developer conference, Huang revealed a roadmap for AI, setting out where the technology may be heading and the potential use cases. Moving from generative AI, which is currently used for content creation, the next stage is likely to be agentic AI, offering customer service and patient care. The final destination is physical AI, which involves autonomous vehicles and general robotics.
As the innovation continues, this roadmap is likely to see demand for compute continue to increase. According to Nvidia, data centre capex will exceed $1 trillion by 2028 (from under $300 million in 2023), which we think is plausible5.
Semiconductor valuations
In light of the favourable comments about capex and future demand for AI semiconductors, we believe that valuations for semiconductor firms look the most attractive they’ve been since the end of 2022. For example, Nvidia was recently trading below valuation levels seen in 2020 and 2022, although it has risen following the US-China trade truce.
Our optimism about the sector also stems from the fact that investor sentiment is extremely negative. This is often a precursor to a “bottom” in the market cycle. Indeed, we find it encouraging that some companies in this sector which have struggled finally appear to have reached the bottom and are seeing their fortunes improve.
For example, Texas Instruments, which is a semiconductor bellwether and the poster child for the semiconductor cycle, reported better-than-expected earnings and raised its guidance. In the earnings call, management said there was more evidence and signals that the market and their business was recovering6. Microchip provided similar commentary on its call.
We are also encouraged by the sector’s robust fundamentals and the prospect that demand for compute could continue to grow in the coming years, driven by ongoing innovation.
Given our belief in the AI roadmap and trajectory of compute, we felt the best way to take advantage of the recent dislocation in the marketplace was to add to our positions in AI chip leaders Broadcom and Nvidia. We have also increased our holdings in semiconductor-equipment players, such as ASML and Applied Materials, that we think will benefit from the expected robust capex trends ahead.
1CNBC, ‘Microsoft pops more than 7%, posts best day since 2022’, (cnbc.com), May 2025.
2PPC Land, ‘Meta’s ad business soars as AI tools drive 16% revenue growth in Q1’, (ppc.land), May 2025.
3Meta Press Release, ‘Meta Reports First Quarter 2025 Results’, (investor.atmeta.com), April 2025.
4Tech Monitor, ‘Alphabet shrugs off tariff uncertainty and proceeds with $75bn capital expenditure plan’, (techmonitor.ai), April 2025.
5Nvidia, ‘GTC 2025 Keynote Deck’, (nvidianews.nvidia.com), March 2025.
6Refinitiv StreetEvents, ‘Q1 2025 Texas Instruments Inc Earnings Call’, (investor.ti.com), April 2025.
The value of investments will fluctuate, which will cause prices to fall as well as rise and investors may not get back the original amount they invested. Past performance is not a guide to future performance. The views expressed in this document should not be taken as a recommendation, advice or forecast, nor a recommendation to purchase or sell any particular security.
