Big Tech’s "Miracle on Costs": Are Wall Street’s AI Assumptions Too Good to Be True?

Wall Street’s current narrative for Big Tech hyperscalers—Alphabet, Amazon, Meta, Microsoft, and Oracle—requires a massive leap of faith. Analysts are betting these tech giants can scale AI revenue exponentially while simultaneously shrinking their operational overhead.

However, looking under the hood of these financial models reveals assumptions that stretch historical precedent to its limit.

The Core Bull Case: Unprecedented Margin Expansion

To justify current stock valuations, consensus analyst models rely on a dramatic widening of operating margins over the next six years:

Operating Margins: Projected to jump from 27% in 2023 to ~31% by 2029 across the five hyperscalers—the highest group level since Meta went public in 2012.

The SG&A Drop: Roughly half of this margin expansion relies on Sales, General, and Administrative (SG&A) costs dropping from 10% of revenue down to ~8% by 2029.

The Dollar Scale: Because combined annual revenue is expected to top $3 trillion, a mere 2% drop in SG&A represents $77 billion in reduced corporate spending in 2029 alone.

The Accounting Reality: Capex vs. Fixed Depreciation

While analysts model falling overhead, Big Tech is pouring trillions into AI data centers, chips, and physical infrastructure. Under standard accounting principles (GAAP), that capital expenditure translates into unavoidable fixed costs:

Soaring Depreciation (D&A): As new data centers go live, Depreciation & Amortization expenses will surge as a percentage of revenue.

No Room for Error: Because depreciation costs are locked in once infrastructure is built, analysts are using aggressive SG&A cuts as a "plug" to keep operating margins expanding.

[ Massive AI Capex Spent Today ]

│

▼

[ Higher Fixed Depreciation (D&A) ]

│

▼

Must be offset by [ Deep SG&A Expense Cuts ]

to achieve [ Target Operating Margins ]

The Skeptic’s View: Is AI Productivity Being Overpromised?

The entire financial narrative rests on a single premise: Big Tech will use its own internal AI tools to achieve unprecedented workforce efficiency.

"Whenever growth is this fast, there's got to be some strong positive correlation between SG&A and revenue... I can't think of a scenario where that's ever happened before."

— Kevin Koharki, Purdue University Accounting Professor

Historically, scaling revenue requires hiring more sales reps, expanding customer support, and growing administrative staff. If internal AI productivity tools fall short of these ambitious projections, overhead will scale alongside revenue, and projected operating margins will crumble.

Balance Sheets Are Already Feeling the Strain

This isn't just a paper accounting debate; SG&A consumes real cash for payroll, marketing, and sales commissions. The massive AI buildout is already impacting cash flows across the sector:

Negative Free Cash Flow: Both Amazon and Oracle have recently dipped into negative free cash flow territory due to heavy capex spending.

Equity & Debt Financing: Alphabet halted stock buybacks to issue equity to fund spending, while credit agencies have downgraded Oracle's credit rating closer to junk status.

Wall Street analysts appear to be reverse-engineering their expense forecasts to match management guidance rather than modeling historical realities. For Big Tech to hit these price targets, AI cannot just be a new revenue stream—it must deliver a workplace productivity miracle that offsets hundreds of billions in physical infrastructure costs.

-Chart from WSJ