"The 'AI War' cannot be lost! If the Mag 7 stocks on the US market burn their cash flow into negative territory this year, what does that mean for the market?"

"The 'AI War' cannot be lost! If the Mag 7 stocks on the US market burn their cash flow into negative territory this year, what does that mean for the market?"

When the AI capital expenditures of Silicon Valley giants balloon to nearly the scale of their entire annual cash flow, the market’s concern is no longer “is it worth it,” but “can they sustain it.” According to the latest public data, Google, Amazon, Microsoft, and Meta—the four hyperscale cloud firms—have combined capital expenditure guidance for 2026 at about **$650 billion**. Including Oracle and CoreWeave, the total rises to **$740 billion**. These numbers are not only above market expectations, but represent a **multiplicative deviation**. What does **$740 billion** mean? - $740 billion is about a **70% year-on-year increase** from 2025; - It is **twice** the market’s consensus expectation (about 35% capex growth) by the end of 2025; - $740 billion is close to the **total annual operating cash flow** of the entire hyperscale cloud system. - **More worryingly, as Goldman Sachs analyst Shreeti Kapa points out, if spending reaches this level, its intensity will approach 1.4% of GDP—the peak of the dot-com bubble in the 1990s.** While still less than the intensity of the Industrial Revolution, it's already rare in modern tech history. ZeroHedge, a well-known financial blog, wrote: > “These numbers are so huge that we immediately joked: after using all their free cash flow to pay for capex, the Mag 7 in 2026 (and perhaps longer) will be unable to afford any stock buybacks at all.” What really shakes the market is not a single company “overspending,” but the **entire hyperscale cloud system’s capital expenditure simultaneously running out of control**. This is not an ordinary capex bump, it's a **structural leap**. ## Cash Flow Is Being “Devoured” by AI There’s not enough money for stock buybacks—it’s no longer a joke, but an unfolding reality. Goldman Sachs estimates that if 2026 capex reaches the $700 billion level, that number would **almost equal all operating cash flow of hyperscale cloud firms**. Bank of America’s more detailed model concludes: - **Microsoft’s solitude:** In 2026 **only Microsoft** is expected to have operating cash flow covering capex. - **Meta’s turning point:** Meta has hinted it could at some point move from **“net debt neutral” to “net debt positive.”** - **Others:** Even if buybacks stop completely, free cash flow will be drained. Bank of America wrote: > “Except Microsoft, even slowing or skipping stock buybacks in Q4, other firms’ cash flow surpluses will shrink significantly.” This means if capex keeps rising, cash balances will drop quickly, **debt financing will be unavoidable—which will become a big problem.** ## AI Turning Into a Debt Bubble: Related Debt Makes Up 14% of U.S. IG Bond Market As internal cash flow is no longer enough to cover spending, tech giants are forced into the bond market on a large scale. A few months ago, ZeroHedge warned: “AI is now also a debt bubble, quietly overtaking all banks to become the largest segment in the market.” Bloomberg wrote in the latest “Credit Weekly”: > “Big tech firms are preparing to spend far more on AI than investors previously expected, and whatever the outcome, fund managers are increasingly worried that the credit market could be hit.” Just in the week leading to February 11, 2026, the market witnessed a frenzy: - **Oracle:** Issued a record $25 billion in bonds. Even as its stock plunged due to negative cash flow and soaring default risk, the bond issue drew $129 billion in orders. - **Google:** Following Oracle, Google issued $20 billion in dollar bonds (original plan $15 billion)—its largest ever, with over $100 billion in orders. Google even plans to issue a **100-year bond**, the first such attempt by a tech company since the dot-com bubble. Why so much debt? Because revenue from ads and cloud services is nowhere near enough. Estimates say **global data centers will need about $2.9 trillion in capex before 2028** (and counting), while firms’ own operating cash flows can only cover about half. How to fill the remaining **$1.5 trillion** gap? The answer is only one: **debt**. This includes corporate bonds, asset-backed securities (ABS/CMBS), private credit, and even sovereign debt. **By the end of 2025, AI-related investment grade debt makes up 14% of the U.S. IG market, becoming the single largest thematic segment—bigger than the banking sector.** Morgan Stanley expects the investment grade bond issuance in tech, media, and telecom may reach **$2.25 trillion** in 2026, a record high. ## Cracks Have Appeared in the Bond Market Despite strong demand, cracks are starting to show. Bloomberg data shows U.S. investment grade corporate bond spreads widened by **2 basis points** last week. Oracle’s new $25 billion bond noticeably underperformed Treasuries in secondary trading. After Oracle announced selling stock to raise cash, market anxiety surged and its stock plummeted. F/m Investments CEO Alexander Morris said: > “The investment boom in AI has indeed attracted many buyers, but the upside is limited and the margin for error is extremely tiny. **No asset class is immune to depreciation.**” Current equilibrium is extremely fragile. The market is on “autopilot”—so long as the AI story continues, the bond market doors stay open. But a shock like the “DeepSeek moment” in January 2025, or major disruption to tech giants’ moats, could slam the bond market shut instantly. ## Chain Reaction in the Software Industry and Private Credit AI isn’t just draining giants’ cash flow—it’s shattering traditional software valuation logic, laying the biggest mine for the credit market. Bloomberg notes that as AI tools increasingly penetrate professional services, investors are repricing the growth outlook for the entire software sector. - **AI boosts efficiency = falling software demand:** As firms like Anthropic roll out AI tools for professional services, investors worry that AI will obsolete many SaaS products. If AI writes code or creates reports, why buy expensive software licenses? - **Software company bonds are being dumped:** Software company leveraged loan prices are down about 4% this year. - **Private credit gets hit:** This is the riskiest link. According to Barclays, the software sector is the largest risk exposure for Business Development Companies (BDCs—listed private credit funds), making up **20%** of their portfolios. Barclays report notes: > “Software is the largest sector exposure in BDC portfolios, about 20%, making the sector particularly sensitive to the recent drop in software equity and credit valuations.” When AI giants burn cash building infrastructure, they’re actually creating technology that could kill their downstream customers (software firms). If software companies default because their products are replaced by AI, holders of their heavy debts—private credit markets—will collapse first, triggering a chain reaction. ## Prisoner’s Dilemma: Why Bet Even Knowing It's a Bubble? Facing Goldman Sachs’ doubts about “too much input, too little output,” why do CEOs at Google, Microsoft, Amazon still choose “full speed ahead”? **The answer lies in game theory’s “Nash equilibrium.”** For the giants, it’s a classic binary strategic choice: - **If you don’t invest:** Permanently losing market share. AI infrastructure has a “winner-takes-all” dynamic. Fall behind now, you never catch up. Like IBM missing cloud computing, facing strategic obsolescence. - **If you invest but overdo it:** Financial statements suffer, margins get squeezed, returns stretch due to overcapacity. But at least you survive—and stay at the table. - **Prisoner’s dilemma:** If rivals invest and you don’t, you lose customers; if you invest and rivals don’t, you win market share. Thus, **the rational strategy is always to invest**. As Goldman Sachs analyzed, this dynamic creates a Nash equilibrium: even with compressed near-term returns, continued capex is rational at the individual level. This is why, even at risk of moving from “net cash” to “net debt,” and bearing hundreds of billions in debt, the giants won't stop. Because for them, **shrinking market cap (from worsening finances) is tolerable, but disappearance (from falling behind in tech) is intolerable.** ## Endgame Simulation: Trillion-Dollar Profits or Ruin? The end of all this depends on one core question: **Return on Investment (ROI).** Goldman Sachs analyst Shreeti Kapa calculated: > Over the past decade, big tech profits have been typically 2-3 times their capex. Given average capital expenditures of $500–600 billion per year from 2025–2027, to maintain investors’ usual returns, these firms would need to **achieve more than $1 trillion in annual profit run rate**. But the consensus estimate for 2026 profits is only $450 billion. That's a huge gap. Even the most optimistic strategists struggle to explain how $30/month subscriptions and occasional enterprise contracts can double tech giants’ profits in the short term. Goldman gives two possible outcomes: 1. **Bull scenario (Cloud 2.0):** AI adoption follows the path of cloud computing. AWS broke even in 3 years, achieved 30% operating margin in a decade. If AI repeats this, today’s massive capex returns astonishing profits. The current $1.5 trillion cloud backlog supports this narrative. 2. **Bear scenario (Global Crossing redux):** History shows pioneers of major tech shifts often die on the beach (e.g., Global Crossing in the fiber era). Today’s giants are richer, but current spending and intensifying competition suggest **not all big firms will generate enough long-term profits to reward today’s investors.** Before this grand bet is decided, the bond market “vigilantes” may wake up first. If they decide to stop underwriting this feast, this debt-fueled AI boom could end in a dramatic way. #### Risk warning and disclaimer The market carries risks, and investment requires caution. This article does not constitute personal investment advice, nor does it take into account the unique investment objectives, financial circumstances, or needs of individuals. Readers should consider whether any opinions, views, or conclusions in the article are appropriate for their specific situations. Investments made accordingly are at your own risk.