Arthur Mensch argues that the core knowledge for training advanced AI models is limited and circulates quickly among top labs. This diffusion of knowledge prevents any single company from creating a sustainable IP-based lead, which is accelerating performance convergence and commoditization across the industry.

As AI models democratize access to information and analysis, traditional data advantages will disappear. The only durable competitive advantage will be an organization's ability to learn and adapt. The speed of the "breakthrough -> implementation -> behavior change" loop will separate winners from losers.

An often overlooked indicator of national competitiveness in quantum is 'cycle time'—the duration from idea to testable prototype. While the US excels at research, long fabrication lead times (e.g., 18 months for a photonic circuit) create a major disadvantage compared to regions where it takes weeks, hindering the rate of innovation.

A "software-only singularity," where AI recursively improves itself, is unlikely. Progress is fundamentally tied to large-scale, costly physical experiments (i.e., compute). The massive spending on experimental compute over pure researcher salaries indicates that physical experimentation, not just algorithms, remains the primary driver of breakthroughs.

Unlike software distributed instantly through browsers, physical AI diffuses slowly across varied industries, geographies, and machines. This makes time and longevity critical factors. Customers need a stable, long-term partner, making it difficult for new, less-established startups to compete.

New chip fab ventures face immense hurdles because fabrication is less like following a manual and more like mastering a recipe through decades of trial and error. This accumulated, non-transferable knowledge, likened to "cooking," creates a significant moat for incumbents like TSMC.

In a world where AI implementation is becoming cheaper, the real competitive advantage isn't speed or features. It's the accumulated knowledge gained through the difficult, iterative process of building and learning. This "pain" of figuring out what truly works for a specific problem becomes a durable moat.

Unlike semiconductors, where the U.S. has a substantial lead, quantum is a new field where the competitive moat is small. This creates a thin margin for error in industrial policy and R&D strategy, demanding a higher degree of precision from the outset.

Unlike the AI industry, which requires massive capital investment, quantum computing allows Britain to compete effectively with larger economies like the U.S. This lower financial barrier to entry leverages Britain's strong research base, making it a uniquely competitive player in the emerging quantum sector.