Silicon railroads: How the $600 billion AI capex pays for itself

The financial headlines of 2026 have been dominated by a single, staggering figure: $600 billion. That is the projected aggregate capital expenditure of the US Hyperscalers; Microsoft, Alphabet, Meta, and Amazon. To the casual observer, this looks like an arms race with no ceasefire in sight. To the sceptic, it looks like the over investment calamities of 1999 returning to haunt US Technology.

When you look at the underlying capital efficiency, a different story emerges. This isn't a speculative bubble, it is the most well funded infrastructure build out in recent times. Managed by companies with a proven capital efficiency guardrail. Superior Return on Equity (ROE) and Return on Invested Capital (ROIC).

The Quality Guardrail: ROE vs The Spend

The primary fear is that massive capex will lead to margin shredding, large amounts of indebtedness and a sharp halt to EPS growth. Historically, when a company increases its investment in physical assets, its return ratios typically compress. Yet, the US technology giants continue to defy gravity.

If we look at the Mega Cap names, the average Return on Equity (ROE) remains well above 30%, compared to the S&P 500 average of approximately 19%.

Table 1 – Return on Equity of the largest 7 stocks in the S&P 500 vs S&P 500 average

Source: Bloomberg February 2026

Chart 1 - 5-Year Trend of Aggregate Capex vs. ROE for Mag 7

Source: Bloomberg February 2026

*2026 representing the $600bn Capex figure

This divergence is critical. For every dollar these companies reinvest, they are generating a return far exceeding their cost of capital. They aren't just spending, they are compounding.

This is Not 1999: The Self Funding Reality

The comparison to the dot com era fiber optic glut falls apart under the weight of the balance sheets. In 1999, tech spending was largely fuelled by external debt. In 2026, this capex cycle is being backed almost entirely by operating cash flow.

Chart 2 – The indebtedness of Mega Cap companies over the last 5 years

Source: Bloomberg February 2026. Where Net Debt is the sum of the company's liabilities and debts with its cash and other similar liquid assets

Even after accounting for these historic investment levels, the Free Cash Flow (FCF) margins of the top tier US tech names remain robust. They are building the railroads using their own assets, not borrowing capital it cannot repay. Furthermore, unlike the 90s telecommunications, who built pipes for others to use, these firms are the primary customers of their own infrastructure. Meta uses its chips to refine its own advertising stack, Microsoft uses its data centres to power its own software-as-a-service. The utilisation is internalised.

ROIC: The Ultimate Management Scorecard

While ROE captures the shareholder view, Return on Invested Capital (ROIC) tells us how efficiently management is deploying all available resources. For a sector shifting from asset light to asset heavy, ROIC is the true source of truth.

Chart 3 - ROIC Comparison: Tech Giants vs. Traditional Industrial & Energy Peers

Source: Bloomberg February 2026. Where ROIC is 100 x (Trailing 12 month Net operating profit after tax / Average invested capital)

These mega cap companies have spent the last decade optimising their margins to the point where they can now afford the hardware phase of the AI revolution without diluting their fundamental quality.

The Harvest Phase: What Comes Next?

There is always a lag between a capex spike and the subsequent reacceleration in earnings. As the 2026 build out matures, we expect to see significant operating leverage. Once the infrastructure is in place, the marginal cost of serving the next AI query or cloud instance drops significantly. Investors who exit now because of capex fears risk missing the high margin tailwinds that follow.

Investing in Capital Efficiency: HUGE, BEST and WWWW

The challenge for investors is no longer identifying the trend, it’s navigating the concentration. While the rewards of this capex cycle are immense, so is the execution risk in the build out.

Investors looking to capture this structural shift can gain diversified exposure through our US focused ETFs:

• ASX: BEST : This ETF is designed for investors seeking exposure to companies that exhibit high levels Return on Invested Capital and Free Cash Flow Margin.
• ASX: HUGE : Designed for those who want exposure to the global leaders with resilient ROE and cash flow profiles. It targets the Hyperscalers whose capital discipline remains a core pillar of their growth strategy.
• ASX: WWWW : Provides exposure to the broader technology ecosystem, capturing the beneficiaries of that $600 billion spend. From semiconductors to software.

Chart 4 – Total Return of BEST, HUGE, WWWW compared to broad US markets

Source: Bloomberg February 2026
Past performance is not a reliable indicator of future performance

By using an ETF structure, investors gain a basket of stocks that are systematically screened based on pre determined parameters. It allows the investor to participate in the AI cycle without the volatility inherent in backing a single balance sheet.

ETF Shares