AI Capital Spending · Concentration Risk
You Already
Own This Bet
The largest companies in America are spending trillions on artificial-intelligence infrastructure today and hoping the revenue arrives later. If you hold an index fund, you are funding that wager whether you chose to or not.
The setupFor most of the past decade, the defining feature of the technology giants was that they could not spend their own cash fast enough. Microsoft, Alphabet, Amazon and Meta generated profit at a scale with no precedent in corporate history. That era has quietly ended. In 2026 those four companies are on track to spend on the order of $700 billion building out AI data centers, chips and power — and the money is no longer coming entirely from the cash they generate.
Bank of America estimates that hyperscaler capital spending now consumes roughly 94% of the group's operating cash flow after dividends and buybacks, leaving little internal cushion and pushing the companies toward outside financing. This is not a story of weak businesses. It is a story of the most profitable businesses on earth choosing to outspend their own cash generation on a single bet — and turning to debt and equity markets to do it.
Exhibit AGoogle sells stock for the first time since 2005
On June 1, 2026, Alphabet announced an $80 billion equity raise — later upsized and priced at roughly $85 billion — to fund AI compute. It was the company's first equity offering in about two decades. The last time Google issued stock, the iPhone did not exist and it had not yet bought YouTube. The package included a $10 billion private placement from Berkshire Hathaway, a notable vote from a firm built on buying cash-generative businesses, not capital-hungry ones.
The context matters more than the headline. Asked on a recent earnings call what kept him up at night, CEO Sundar Pichai pointed to compute capacity — power, land and supply-chain constraints — as the company's top concern. A business that produces over $100 billion a year in advertising profit had just told the market that internally generated cash was no longer sufficient to keep pace.
Exhibit BAmazon's free cash flow is set to go negative
Amazon expects to spend roughly $200 billion in 2026. Against operating cash flow near $140 billion, that math forces free cash flow negative for the year — Morgan Stanley models a deficit of about $17 billion; Bank of America about $28 billion. In an SEC filing, Amazon advised investors it may raise additional debt and equity as the build-out continues. Its long-term debt has already climbed to roughly $119 billion, up from about $66 billion a year earlier. The most efficient cash machine in retail history is preparing to burn more than it makes.
The 2026 Capital Cycle, By the Numbers
~$700B
Combined 2026 AI capex across the four hyperscalers
94%
Of operating cash flow consumed by capex, per Bank of America
$85B
Alphabet equity raise — first since 2005
~⅓
Of the S&P 500 sits in the Magnificent Seven
The four spendersWhat the build-out is doing to cash
Every one of the four is now leaning on bond markets and, in Alphabet's case, equity markets to keep building. The five largest hyperscalers raised about $108 billion in bonds in 2025 alone. The table below frames the scale of 2026 spending against its effect on cash.
| Company | 2026 capex (approx.) | Effect on free cash flow |
|---|---|---|
| Amazon | ~$200B | Negative: −$17B to −$28B (est.) |
| Alphabet | ~$175–190B | FCF projected to fall ~90%; $85B equity raise |
| Meta | up to ~$135B | FCF projected to fall ~90% (Barclays est.) |
| Microsoft | rising, slower pace | FCF projected ~−28% before recovery (est.) |
Figures are company guidance and Wall Street analyst estimates (Morgan Stanley, Bank of America, Barclays, Pivotal Research) as reported early 2026, and are approximate. Estimates vary by source and will change as the year develops.
The first crackBroadcom tripled its AI revenue — and got punished
Broadcom designs the custom AI chips behind Google, Meta, OpenAI and Anthropic. In its quarter reported in early June, total revenue rose 48% to a record $22.2 billion, and AI-semiconductor revenue more than doubled — up 143% — to $10.8 billion, with guidance for it to roughly triple to $16 billion next quarter. The market's response was to sell the stock roughly 15% in a single session, erasing several hundred billion dollars of value and dragging the entire chip sector down with it.
Three things from CEO Hock Tan's commentary explain the reaction. He acknowledged that Google will draw on multiple chip suppliers rather than Broadcom alone. He signaled a retreat to selling chips only rather than complete integrated AI systems, with the surging AI mix diluting overall margins. And despite the demand, he reiterated rather than raised the long-term target of $100 billion-plus in AI revenue. When a company can triple a business line and still disappoint, expectations have detached from any plausible reality.
Why this is your problem
Apple, Microsoft, Amazon, Alphabet, Meta and Nvidia now account for roughly 30% of the S&P 500 — the full Magnificent Seven, close to a third.
If you own a 401(k) or a broad index fund, you are already positioned in this bet. You did not choose the concentration. The index chose it for you.
Two outcomesThe bill comes first; the revenue is a hope
This is the part that should focus the mind of anyone within a decade of retirement. Every one of these companies tells you AI will generate trillions in revenue. The accounting says they are spending the trillions first. The return is a forecast, not a fact. Two paths sit ahead, and the index does not let you opt out of either.
If the revenue arrives
This becomes the largest productive infrastructure build-out in modern history — larger than the railroads, larger than the internet — and the spending is vindicated.
If it doesn't
Companies making up roughly a third of the U.S. market have leveraged their balance sheets into the largest write-down cycle since 2000 — and this time they are the backbone of the economy, not fringe start-ups.
The fiduciary point
None of this requires predicting which outcome wins. It requires knowing that your retirement is exposed to the answer — and deciding, deliberately, how much of that exposure you are willing to carry.
Concentration risk is not a forecast. It is a position you already hold.
This commentary is provided by Bailey Financial Services, Inc. for educational purposes only and reflects publicly reported figures and analyst estimates as of June 2026. It is not investment, legal or tax advice, and it is not a recommendation to buy or sell any security. Capital-expenditure plans and free-cash-flow estimates are company guidance and third-party analyst projections that are inherently uncertain and subject to change. Index weightings fluctuate. Past performance does not guarantee future results. Individual circumstances vary; consult an advisor before acting.
A Conversation, Not a Sales Pitch
Do you know how concentrated your retirement actually is?
If a meaningful share of your portfolio rests on a single thesis playing out, it is worth understanding before the market decides for you. We help retirees and pre-retirees measure and manage concentration and sequence risk — as a fee-only fiduciary.