John Ian Lau

Contributing Editor

Solomon Grey Capital

To our 22 million readers: Welcome to another edition of Solomon Grey Capital, where we cut through the noise to spotlight the trends that matter most for your portfolio.

A New Secular Force Has Arrived

History rarely announces its turning points in real time, but we believe we are living through one. Artificial intelligence is not another cyclical tech theme—it is the fourth great general-purpose technology of the modern era, alongside the steam engine, electricity, and the internet. Like those predecessors, AI will restructure productivity, corporate profit pools, and the very nature of work over the next two decades.

The numbers are already staggering:

• Global data centre power demand is forecast to double by 2030, with AI accounting for ~70% of the increase (IEA).

• Capital expenditure on AI infrastructure from the Magnificent 7 alone is expected to exceed $300 billion in 2026 (Goldman Sachs).

• McKinsey estimates AI could add $13–25 trillion to global GDP by 2040—roughly the size of the U.S. and China economies combined today.

• Productivity impact: Goldman Sachs now projects U.S. GDP growth 1.5 percentage points higher per annum over the next decade purely from AI adoption.

Even Cathie Wood, rarely accused of understatement, recently raised her five-year revenue forecast for Tesla’s robotaxi business alone to $1 trillion. Coatue’s Philippe Laffont, speaking at their 2025 CEO Summit, went further: “We are still in the 1994–95 phase of the internet. The real value creation—the applications layer—hasn’t even started.”

Eight Decades of Market History Meet the Greatest Capex Cycle Ever

One of the most enduring truths in markets remains intact: bull markets last far longer—and deliver far greater cumulative returns—than bear markets. Since 1949, the average S&P 500 bull has lasted 5.3 years and returned +254%. The average bear? Just 11 months and –31%.

Yet every great technology wave compresses and then dramatically re-extends that cycle.

The internet era is the clearest parallel. From 1995 to 2000, the Nasdaq rose +580% as infrastructure was laid. A brutal –78% bear market followed in 2000–02. Then came the longest bull market in history—delivering +527% from 2009 to 2021—as the application layer (Google, Facebook, Amazon, Netflix, mobile) finally scaled.

We are now repeating the pattern—only faster and larger.

Phase 1 (2022–2025): The “Picks and Shovels” boom
Nvidia, data-centre power, copper, optical networking, and the hyperscalers have been the primary beneficiaries. Public markets captured the early value, just as Cisco, EMC, and Sun Microsystems did in the late 1990s.

Phase 2 (2026–2035): The Application & Agentic Layer
This is where the real economic surplus will be created—and, critically, where much of the value may accrue to private rather than public companies. The median tech company now reaches IPO at 14 years old with $220 million of revenue—versus 8 years and $44 million in the internet era (University of Florida study). Agentic AI systems, vertical SaaS killers, and autonomous enterprise workflows are being built in private.

As Coatue’s Laffont warned: “If you’re waiting for public market exposure to the true AI winners, you may be waiting a very long time.”

The AI Infrastructure Grand Prix – 2026 Edition

The race to build planetary-scale compute continues to resemble Formula 1 on steroids:

• Oracle (Red Bull): Ignoring the budget cap entirely—$10 billion+ deals with OpenAI and others. Brute force works.

• Microsoft (Alpine): Whipsaw strategy—full send, sudden braking, back to full send. Azure AI capex guidance keeps getting upgraded.

• Google (McLaren): The integrated stack (TPU + Gemini) gives terrifying straight-line speed, but one degree of temperature change (CUDA dependency risk) and the car becomes undriveable.

• CoreWeave (Aston Martin): The new-money neo-cloud—zero legacy debt, infinite VC fuel, 200,000+ Blackwell GPUs on order. Speed is astonishing; reliability still TBD.

• AWS (Ferrari): Finally ditching the home-grown EFA gearbox for standard Nvidia networking. A painful but necessary admission.

What This Means for Portfolios

1 Own the secular enablers: AI infrastructure is a multi-decade capex super-cycle. Power generation, utilities, copper, and specialised semiconductors remain structurally under-owned.

2 Beware narrow leadership: The S&P 500 is now more concentrated than at the 2000 peak. History rhymes.

3 Private markets will matter more than ever: The best AI application companies will stay private longer. Traditional 60/40 investors risk a decade of structural underperformance.

4 Volatility is coming—but it will be relatively brief. The next bear market, whenever it arrives, is likely to be sharp, sentiment-driven, and short—creating the entry point for the application-layer bull that follows.

Final Thought

Markets spend far more time climbing walls of worry than sliding down slopes of hope. AI is the most powerful wall of worry in a generation—expensive, overhyped in the near term, and massively under-appreciated in the long term.

As ever, the greatest risk is not overpaying for the future—it is missing it entirely.

Views expressed are those of Solomon Grey Capital and do not constitute investment advice. Past performance is not indicative of future results.