Scotiabank Sees AI Capex Topping $1 Trillion, Debt Model Shifts

The Portfolio Strategy team at Scotiabank released its August Monthly Chart Book on August 5, 2026, detailing a significant inflection point in market dynamics driven by artificial intelligence investments. The report highlights strong second-quarter earnings, with an aggregate 13% beat and median earnings per share growth of 38%, while flagging that a key metric is inflated by Alphabet's unrealized gains. Most notably, the analysis makes the case that the foundational asset-light business model of U.S. tech is morphing under the weight of AI-related spending, with consensus capital expenditure for the MAG-7 group projected to surpass $1 trillion. This shift is accompanied by a rise in use, with the AI basket's net debt to forward EBITDA climbing to 2.0x when including off-balance-sheet obligations. Alphabet's stock traded at $362.78, down 2.87% on the day, as of 17:18 UTC today.

Context — [why this matters now]

The current market strength is directly attributable to corporate earnings performance. For over 15 years, the dominant U.S. technology sector thrived on an asset-light model characterized by minimal debt and high free cash flow generation. This paradigm is now undergoing a fundamental change as companies commit unprecedented capital to build AI infrastructure. The last comparable shift in tech sector behavior occurred during the dot-com bubble of the late 1990s, when heavy capital expenditure on internet infrastructure preceded a market correction, though the underlying technology and revenue potential of AI are considered more substantial. The catalyst for this analysis is the convergence of peak earnings beats with a tangible rollover in free cash flow as spending accelerates, signaling a new phase in the market cycle where financial discipline is being tested against growth ambition.

Data — [what the numbers show]

Scotiabank's data presents a strong but nuanced earnings picture. For the second quarter, 85% of S&P 500 companies have topped earnings estimates, producing an aggregate beat of 13% and a median beat of 5.9%. Top-line revenue growth is running at 13% year-over-year. The headline EPS growth of 38% is notably boosted by non-operational gains; the strategists specify it is "boosted by Alphabet's unrealized gains on its stakes in Anthropic and SpaceX." The core of the report focuses on AI investment metrics. Consensus forecasts now project capital expenditure for the MAG-7 companies to exceed $1 trillion over the coming 12 months. A critical use metric, which includes off-balance-sheet obligations, shows the AI basket's net debt/forward EBITDA has climbed to 2.0x. In comparison, the S&P 500 forward price-to-earnings ratio sits at a premium to the TSX, which trades at 15.7x forward earnings, representing a 21% discount versus its five-year median discount of 29%. Alphabet's share price ranged between $356.82 and $384.47 during the trading session.

Metric	MAG-7 / AI Basket	TSX Composite
Forward P/E	~28x (est.)	15.7x
Net Debt/EBITDA	2.0x	N/A
YTD Performance (USD)	Outperforming	+2.3% in July

Analysis — [what it means for markets / sectors / tickers]

The shift toward a capital-intensive model implies higher volatility for mega-cap tech stocks, as Scotiabank frames it, "but not necessarily the end of the trade." The second-order effect is a potential rotation within equity markets. While Scotiabank maintains a slight overweight on U.S. tech due to strong earnings revisions—tech leads S&P 500 forward EPS revisions at +18% over three months—the growing debt burden creates a divergence between well-funded incumbents and speculative challengers. Sectors positioned to benefit from AI infrastructure build-out include semiconductors, networking equipment, and utilities. A key risk to this outlook is the requirement for tangible returns on the massive AI investments; at some point, the spending must translate into profitable new revenue streams, or investor patience will wane. Current market positioning shows flows remaining heavily concentrated in the MAG-7, but the report highlights a speculative chart pairing U.S. software ETFs like IGV against a basket of Canadian tech names, suggesting some investors are seeking value in overlooked areas.

Outlook — [what to watch next]

The immediate catalyst for the AI trade will be third-quarter earnings reports beginning in mid-October, where commentary on capex budgets and AI monetization will be scrutinized. For the U.S. dollar, whose downtrend Scotiabank believes will extend after the DXY fell below 100, the next major data point is the August CPI release on September 12. A weaker dollar setup is positive for gold, which the report notes traded a full standard deviation below its 200-day average; a sustained break above its 50-day moving average near $4,200 would confirm a bullish reversal. Investors should monitor the TSX small-cap index, which is up 51.2% year-over-year but still trades at a discount to its historical average, for signs that the rally is broadening beyond large-cap U.S. tech.

Frequently Asked Questions

What does a 2.0x net debt/EBITDA ratio mean for tech stocks?

A net debt to EBITDA ratio of 2.0x indicates that a company's total debt, net of cash, is twice its annual earnings before interest, taxes, depreciation, and amortization. For tech stocks historically accustomed to negative net cash positions, this represents a significant shift in financial structure. It moves the sector closer to the use profiles of more traditional industrial companies, which could lead to higher borrowing costs and increased sensitivity to interest rate changes, potentially compressing valuation multiples over time if the debt does not generate expected returns.

How does current MAG-7 capex compare to historical tech spending cycles?

The projected $1 trillion in MAG-7 capex over the next year is unprecedented in scale for the technology sector. Historical comparisons are limited, but the closest analog is the telecom and internet infrastructure build-out during the dot-com era. That period also featured soaring capital expenditure, but the absolute amounts were far smaller and involved a wider array of companies. The current concentration of spending among just seven firms highlights the immense barriers to entry in the AI hardware race and the winner-take-most dynamics that markets are anticipating.