Ben Rogoff, Manager of Polar Capital Technology Trust, shares insights on how breakthroughs in AI models, soaring adoption, and record infrastructure investment are accelerating AI’s rise as a general purpose technology.
As disruption begins to impact sectors like software and information services, the team is shifting focus from mega-cap incumbents to infrastructure enablers and real-world AI applications, drawing parallels with the early internet era and emphasising the importance of a long-term view.
Ben Rogoff, Manager of Polar Capital Technology Trust: “2025 has been a landmark year for artificial intelligence, marked by rapid advancements in model capabilities and accelerating adoption. New releases from OpenAI, Anthropic, Alphabet, and xAI have pushed the boundaries of AI performance, with innovations in reasoning models and test-time compute enabling more sophisticated outputs. Elon Musk’s xAI claimed Grok 4 as the world’s smartest AI, a title quickly contested by OpenAI’s GPT-5, which now leads on several benchmarks. This competitive momentum underscores the pace of AI progress, which is scaling five to six times faster than Moore’s Law.
Adoption is surging. OpenAI’s user base has grown from 300 million to 800 million weekly users in 2025 alone, now handling over a billion daily queries, just three years after ChatGPT’s launch. For comparison, Google took 11 years to reach a similar milestone. This explosive growth is mirrored in other areas: Cursor users generate a billion lines of code daily, and Microsoft reports that 35% of new software code is AI-generated. Token usage, a measure of AI activity, has skyrocketed, with Google processing nearly a quadrillion tokens in June, more than double the previous month.
These trends are translating into substantial revenue growth. OpenAI’s run rate has jumped from $1 billion in 2023 to $12 billion, while Anthropic is approaching $4 billion. AI is also contributing meaningfully to Microsoft Azure’s growth, and Meta reports AI-driven improvements in ad conversions and user engagement.
Investment is following suit. Cloud giants spent $130 billion on AI infrastructure in 2023, rising to $225 billion in 2024, and are projected to exceed $375 billion in 2025. Sovereign AI spending now surpasses $50 billion annually. Despite this, the industry remains capacity-constrained, with analysts forecasting over $1 trillion in additional infrastructure investment between 2025 and 2027. This scale of spending, potentially 1% of global GDP by 2026, reflects growing confidence in AI’s transformative potential.
This investment boom is benefiting a wide array of companies across compute, storage, networking, and increasingly, power and cooling. As AI models grow more complex, energy demands are soaring. Anthropic estimates that AI training power requirements could reach 20–25 gigawatts by 2028. For context, Microsoft added just two gigawatts of datacentre capacity globally in the past year. We see significant opportunities in sectors supporting AI infrastructure, particularly those addressing energy and thermal management challenges.
Looking ahead, the next frontier is agentic AI, systems capable of autonomous planning and decision-making. This concept, the theme of Polar Capital Technology Trust’s 2025 Annual Report, could revolutionise productivity by overcoming cognitive and time constraints. NVIDIA CEO Jensen Huang envisions a future where 50,000 employees are supported by 100 million AI assistants, unlocking vast efficiencies and new possibilities.
Beyond agentic AI lies artificial general intelligence (AGI), or superintelligence. While timelines vary, leaders like Sam Altman and Mark Zuckerberg suggest it could arrive within a few thousand days. Elon Musk goes further, predicting a near-certain emergence of superintelligence by 2030.
We have positioned our portfolios to capture this shift, pivoting toward AI more than two years ago and embracing the role of “AI maximalists”. While early AI disruption has been modest, recent developments suggest a shift from coexistence with legacy technologies to direct competition. Sectors like executive search, advertising, and software are already feeling the impact.
Software, in particular, faces challenges. AI threatens traditional cloud-based business models and the value of legacy codebases. As AI-generated code becomes more prevalent, the perceived worth of decades-old software diminishes. This disruption is expected to accelerate, creating new market opportunities, what we call “invisible markets”, that may surpass existing ones in scale.
We are adjusting our exposure accordingly, reducing holdings in some of the Mag 7 tech giants and focusing on AI infrastructure enablers, beneficiaries, and companies at the intersection of technology and the real world, such as Axon in law enforcement.
While volatility is expected, we draw parallels to the mid-1990s internet build-out, a period of strong returns despite frequent corrections. We remain focused on the long-term potential of AI as the next general purpose technology, confident in its multi-cycle investment experience and dedicated AI fund to navigate and capitalise on this transformative era.”
