Are the Big Four Tech Names Dead, or Just Resting? The Lifeline of the AI Chain Gets Settled Early Next Year
US Stocks · Options · News · Views
The four hyperscalers, GOOG, AMZN, MSFT, and META, are the real buyers in the AI supply chain. Right now they are squarely "the side that's spending." AI does bring them revenue, but far short of covering the spend, and free cash flow keeps going negative.
This is the lifeline question, because their success or failure decides how far the whole AI thesis runs. The chain is clear: only once these four make money on AI can they keep buying downstream into semis and software. Today's high expectations and high prices downstream are all underwritten by these four's capex forecasts. Once they buckle and stop piling on compute and capex expectations get rewritten, what happens to the downstream names that fed on that spending? Can they ever go back to where they came from?
Worst case: if Big Tech really breaks, it means stopping the overspend, running negative free cash flow for a year or two, then reverting to the core business. They go back. But the suppliers that fed on those high expectations may not revisit today's highs for years. That is the unavoidable source question when looking at semis and AI, and the one most easily ignored when money is crowded in.
Google dropped five full points today, sparked by a DeepMind data scientist leaving for rival Anthropic. But the high-volume drop traces back to positioning and valuation: GOOG has been slowly sliding from its prior breakdown and today slipped to 348. Below that, the 311 to 325 band is the opportunity zone to watch. Forward PE is around 24x, above the S&P's ~20x, but not rich versus its own history. Next earnings will adjust 2026/2027 estimates (2026 raised to double digits, 2027 cut to single digits); the 5-year change isn't large, even slightly positive.
Amazon shows no stabilization pattern today and is still seeking support: first tier 227 to 240, second tier 219 to 226. Below 219 the technical risk expands materially. Forward PE around 25x, one of the heaviest valuation pressures in the group. Price sits slightly above median ($230 spot vs. $212 median), but it's not overvalued.
Microsoft's forward PE is around 19x, below the S&P and below its own history. It is currently below 375, waiting for stabilization anywhere in the zone. META is the most undervalued of the four, forward PE down to about 16x. Price is touching three dense strong-support clusters below: the closer to 500, the higher the margin of safety and the prospective excess return. That is the cadence you control for scaling in.
One flow-data point worth remembering: Jefferies tracked its own trading clients' net selling of the Magnificent Seven over the trailing 20 days at about $6 billion, the second-worst 20-day selling stretch in two years (worst was the $8 billion during the March 2024 selloff). That means a severe sell imbalance has built up, and buyers can step in to rebalance at any time. Near-term selling exhaustion is close to a top: "the big selling is mostly done." But selling exhaustion is not the same as no selling at all.
The real fundamental verdict comes early next year, when Q4 earnings season settles whether AI monetization works. If they fail, the whole AI industry fails, and that old line comes true: today's highs may be unreachable for many years. The Big Four themselves are the most likely to revert. Their core businesses are stable enough; if they just spend less and keep the profits, earnings can keep growing.
The "chop" thesis for the next two weeks now has harder data behind it, courtesy of Citadel Securities, the largest U.S. market maker (handling 20%+ of U.S. equity volume, 35% of retail flow, NYSE market-making for 62% of listed names, 30% of options market-making). One: funding ratios for the top 100 U.S. pensions have reached 110%, the healthiest level since 2001. The better-funded a plan, the more it wants to de-risk. Combined with the classic 60/40 stock-bond allocation, month-end will mechanically see sell-stocks/buy-bonds rebalancing, entirely unrelated to fundamentals.
Two: that negative flow reverses on July 1. Pension contributions, target-date funds, passive allocators, and systematic strategies all re-engage for the second-half campaign. There is outflow, but there are also a lot of institutions that want in. Seasonal probabilities lean bullish: SPX's first-half-of-July hit rate is 69% since 1928, and NDX has been up 17 of the past 18 years (76%). That is not 100%, just the high-probability historical script. Whether this year goes the other way, we watch as we go.
Three: the dip-buying brake light sits around VIX 30. With VIX still below 20, everyone keeps buying. Once a multi-day decline pushes VIX to 30, habitual dip-buyers get scared and step back, and that is often how bottoms get stamped.
Breadth is the first warning sign. The percentage of stocks above their 50-day moving average has slipped from ~60% to 54%, yet the index still grinds higher. This kind of divergence usually resolves the wrong way. Citadel's language isn't that direct, but the substance is: a pullback, a real drop. Market concentration makes it worse. The S&P's top 10 names now make up nearly 40% of index weight; buy $1 of the SPY and ~40 cents flows into the top 10. As trillion-cap names proliferate, the only way the index avoids a material drawdown is for these mega-cap leaders to hold up. If they wobble, nothing else holds up either.
The second weight is the 10-year Treasury yield. One market view holds that, with the Fed withdrawing forward guidance, Treasury investors demand more risk premium, implicitly pushing long-end yields higher. The 10-year is the global benchmark rate; in DCF models, that term amplifies and discounts overall valuations. Growth stocks, the most sensitive to the discount rate, take the most direct hit. How much yields can press valuations depends on how high they run, but directionally this is unambiguously bad for tech.
IBM shot to 265 after-hours on two executive orders: one requiring deployment of a research-grade quantum computer at a national lab by 2028 (the 2028 timing, just before the election, is pointed), the other requiring transition to post-quantum cryptography by 2031. Setting quantum aside, IBM's fundamentals at 250 are reasonable, neither rich nor cheap. Today it reclaimed formerly broken support, building a potential short-term inverted reversal to the upside; overhead, only the 295 to 313 band stands between it and new highs. Speculative money can decide for itself. That said, history is mostly a graveyard of government-corporate partnerships. INTC's new CEO becoming "besties" with the White House is the modern example of "mouth moves, price follows." Treat it as short-term trading and you are calibrated right.
Other quantum names are mostly waiting until 2030 or later for commercialization. IBM itself doesn't expect fault-tolerant machines until 2029, and whether 2028 can be delivered is in doubt. But as long as there is an executive order and heat, money will come to speculate.
SpaceX is back near 150; the last speculative hand has mostly played out. Today's turnover was 30%, a transition phase of chip exchange, waiting for the next speculative cohort to enter.
Disclaimer: This article is personal market observation and trading review only, not investment advice. Markets carry risk; trade with caution.