The money flows¶

Understanding where money actually goes in the technology industry requires distinguishing between impressive announcements and the quieter movements of capital that reveal what informed parties genuinely believe versus what they say at conferences. The hype flows freely because it costs nothing and occasionally attracts additional investment. The actual money flows more cautiously toward opportunities that might plausibly generate returns, which is a considerably more selective criterion than “sounds impressive during pitch meetings.”

Venture capital has discovered that putting “AI” in a startup’s description increases valuation by amounts that would make traditional financial analysts question their career choices. This has predictable effects on startup naming conventions, pitch deck contents, and the number of companies that suddenly discovered their product was AI-powered all along despite previously describing it as sophisticated algorithms. The investors are complicit in this because admitting that “AI” is sometimes marketing rather than meaningful technical distinction would undermine valuations they have already committed to.

The serious money¶

The serious money, meaning amounts that move markets rather than merely fund another cohort of optimistic founders, is flowing primarily into infrastructure. Cloud providers building data centres, semiconductor manufacturers expanding fabrication capacity, and networking companies supplying connectivity represent capital deployment measured in tens of billions of euros annually. These are bets on continued technology growth measured in years rather than quarters.

Cloud providers are collectively investing over €100 billion annually in data centre expansion. These investments have multi-year payback periods and are premised on sustained demand growth that might or might not materialise at projected rates. The investments are too large to quickly reverse. In early 2026, Azure growth hit what analysts described as an infrastructure wall, and Microsoft’s shares fell as the scale of capital investment began visibly weighing on margins. This is either a temporary signal or an early answer to the question of whether the investments were confident positioning or very expensive overconfidence. The Patrician notes that early answers to this type of question are rarely wrong about the direction, only about the magnitude.

Semiconductor manufacturers are expanding fabrication capacity with investments that dwarf cloud infrastructure spending. TSMC’s capital expenditure exceeds €30 billion annually. Samsung, Intel, and others are making comparable commitments. These are decade-long bets on continued semiconductor demand across AI, computing, automotive, and industrial applications. The people making these bets are not guessing blindly. They are guessing with better information than most. They are still guessing.

Venture capital and its enthusiasms¶

Foundation model companies, building large language models and equivalent capabilities, attract the largest venture investments because they require enormous capital for training before generating revenue. The investments require believing that AI capabilities will continue improving, that the companies will successfully commercialise their technology, and that they will not be disrupted by competitors with better technology or lower costs. These are three separate bets, each of which must succeed for the investment to generate returns.

Application layer startups, building products using foundation models from others, receive smaller investments but face questions about defensibility. If your core technology comes from an external provider who might offer competing products, what exactly is the moat? Some companies answer this compellingly. Others discover the answer is “not much” at an inconvenient moment in their growth trajectory.

The venture capital enthusiasm for AI is self-reinforcing in ways that do not obviously end well. Success stories justify additional investment, which attracts more startups, which increases competition, which pressures everyone to raise more capital to outlast competitors. This pattern is familiar from previous technology cycles. It typically resolves through consolidation, quiet failures, and recalibration of expectations. The recalibration is rarely pleasant for those being recalibrated.

Where the money is not going¶

The hype-reality gap is substantial and revealing. Hype concentrates on consumer-facing AI applications and impressive demonstrations. Capital concentrates on infrastructure and enterprise applications with clear revenue potential.

Consumer AI applications generate enthusiasm and user engagement but struggle with business models that justify their valuations. Many impressive demonstrations are loss-making at scale because inference costs exceed revenue per user. Enterprise AI applications receive less hype and more actual deployment, but the deployment has not yet reliably translated into return. An MIT study found that ninety-five percent of companies see zero return on their generative AI investments despite collective spending of thirty to forty billion euros. Enterprises are deploying AI. The productivity improvements that justify the deployment are, in most cases, still being located. This is not fatal to the enterprise thesis, but it is an adjustment to the timeline.

The companies building equipment for AI infrastructure receive modest publicity and substantial capital, because regardless of which AI applications succeed, infrastructure will be needed universally. The suppliers have less risk than the applications companies because they are selling capital equipment rather than renting capacity that must generate returns over years.

The Patrician’s assessment¶

Following the money provides more reliable signals about technology trajectories than following the hype. Large infrastructure investments by established companies signal confidence in sustained demand. Selective venture capital deployment signals investor caution despite public enthusiasm. These signals are not contradictory. They reflect that infrastructure providers and venture investors are making different bets with different time horizons and different risk tolerances.

The Patrician observes that venture returns from current AI investments will likely disappoint because too much capital chased too few genuine opportunities at too high valuations. He observes this without particular satisfaction. The pattern is familiar enough to be predicted and persistent enough that predicting it has no apparent effect on behaviour.

The money flows reveal an industry simultaneously confident enough to deploy enormous capital into infrastructure and cautious enough to require increasingly strong fundamentals before committing to speculative ventures. Whether this reflects rational calibration between opportunity and risk or collective confusion about whether current enthusiasm is justified will be answered over coming years. The answer will be provided by events rather than by analysis, which is historically how these questions are resolved.