According to CNBC, HSBC CEO Georges Elhedery warned at the Global Financial Leaders’ Investment Summit in Hong Kong about a significant mismatch between AI investments and revenues, noting that current revenue profiles may not justify massive spending on computing power. Morgan Stanley estimates global data center capacity will grow six times over five years, with data centers and hardware alone costing $3 trillion by 2028, while McKinsey projects AI-capable data centers will require $5.2 trillion in capital expenditure by 2030. General Atlantic CEO William Ford agreed, describing AI as a “10-, 20-year play” while warning of potential “misallocation of capital” and “irrational exuberance” in early stages. Big Tech firms now collectively expect capital expenditures to exceed $380 billion this year, with OpenAI announcing roughly $1 trillion in infrastructure deals with partners including Nvidia, Oracle and Broadcom.
The Infrastructure Gambit
What we’re witnessing is perhaps the largest infrastructure buildout since the internet’s early days, but with a critical difference: the capital requirements are staggering even by tech industry standards. The McKinsey projection of $5.2 trillion for AI-capable data centers alone represents a bet that demand for compute will materialize at unprecedented scale. Unlike previous technology cycles where software dominated investment, AI’s hardware demands are driving a physical infrastructure boom that requires massive upfront capital with uncertain payback periods. This creates a fundamental tension between the infrastructure providers building capacity and the application layer companies that need to generate sufficient revenue to justify these costs.
The Revenue Reality Gap
The core challenge lies in the timing mismatch that Elhedery identified. While infrastructure spending happens immediately, enterprise adoption cycles for transformative technologies typically take 3-5 years to reach meaningful scale. Most businesses are still in the experimentation phase with AI, running pilot projects and proof-of-concepts rather than deploying mission-critical applications at scale. This creates a dangerous gap where infrastructure providers are building for demand that may not materialize as quickly as anticipated. The Morgan Stanley analysis of sixfold data center growth assumes that enterprise spending will rapidly accelerate to fill this capacity, but history suggests technology adoption follows S-curves rather than straight lines.
Winners and Losers in the Capex Race
Ford’s warning about difficulty picking winners and losers points to a fundamental market dynamic: the early beneficiaries of this spending spree are the infrastructure providers themselves—chip manufacturers like Nvidia, cloud providers like Amazon and Microsoft, and data center operators. These companies are capturing value immediately through equipment sales and service contracts. However, the application layer companies that will ultimately drive AI adoption face a much longer path to revenue generation. This creates a scenario where infrastructure providers may see short-term windfalls while application companies struggle to find sustainable business models. The risk is that we’re building a massive ecosystem without clear evidence that end customers are willing to pay premium prices for AI-enhanced services.
The Productivity Paradox Revisited
We’ve seen this pattern before with previous technological revolutions. The “productivity paradox” of the 1980s and 1990s saw massive IT investments initially fail to deliver measurable productivity gains until business processes were redesigned around new capabilities. AI faces a similar challenge—the technology itself is only part of the equation. Organizations need to fundamentally rethink workflows, retrain staff, and redesign business models to capture AI’s full potential. This organizational transformation takes years, creating the 5-10 year timeline that both executives referenced. The danger is that impatient investors may lose confidence before these long-term benefits materialize, potentially triggering a funding winter that could stall innovation.
Strategic Implications for Investors
The current environment requires a clear-eyed assessment of risk and timeline. Investors need to distinguish between infrastructure plays with near-term revenue visibility and application companies with longer paths to profitability. The “irrational exuberance” Ford mentioned often manifests in valuations that assume rapid adoption and premium pricing that may not materialize. Companies that can demonstrate clear paths to monetization through either enterprise contracts or consumer subscriptions will likely outperform those betting purely on technological capability. The most successful players will be those that can balance ambitious infrastructure investments with pragmatic, incremental revenue generation strategies that bridge the gap until broader adoption accelerates.
