Semis Ripped, but Google Is the Real Switch
US Stocks · Options · News · Views
The indexes closed higher, but the account-level tape is still uncomfortable
All four major indexes closed higher, with Nasdaq and semiconductors leading the rebound. Every member of the Philadelphia Semiconductor Index finished green. On the surface, risk appetite looked alive again. Underneath, the market is still stuck in the same seesaw.
The stronger semis get, the weaker the four big tech names and software look. The index is rising, but breadth has not repaired. The S&P 500 has now logged 53 price-versus-breadth divergences this year, above the 52 seen during the 2000 tech bubble period, and not far from 60 in 2024 or 58 in 2025.
That explains why portfolios can feel strange. The index can look like a bull market while individual holdings feel like a choppy market. A rising index does not mean everything in the account is rising. This is a rotation market, with money moving between a few index weights and a few crowded hot zones.
The real switch is GOOG capex
The most important thing now is not how much semis bounced today. It is the upcoming earnings from GOOG and TSLA. For GOOG, the market will focus on capital spending. Whether capex is raised, maintained, or guided toward slower growth will directly shape the next move in semiconductors.
This semiconductor rebound cannot be separated from AI capex. The second half of 2026 and the fourth quarter are the validation window for AI monetization. If the hyperscalers keep spending on data centers, semis can keep telling the story. If capex shifts from acceleration to slower growth, the market will start compressing valuation again.
That is why a 5% bounce in SOXX does not remove the risk. It only puts the ETF back into a major resistance area. Whether this becomes real repair or just a pressure-zone rebound depends on what GOOG says about future capex.
The bottom in ORCL is not controlled by ORCL alone
ORCL has fallen toward the 120 area, and the obvious question is whether it has bottomed. But this stock can no longer be judged only by its own earnings and technical setup. It is deeply tied to OpenAI, which has become its largest single customer.
The issue is that ORCL borrowed heavily to support the data-center contracts, and the balance sheet pressure has increased. The credit market is already responding through CDS pricing. Bond investors are paying a high cost to insure against default risk, which means credit risk is being repriced.
The contract is worth roughly $300 billion over about five years, beginning in 2027, or close to $60 billion per year on average. OpenAI's own revenue expectations are aggressive: about $3.7 billion in 2024, roughly $13 billion in 2025, $25 billion to $30 billion expected in 2026, and $44 billion to $62 billion expected in 2027.
That creates a hard tension. Even if OpenAI reaches $62 billion of revenue in 2027, that only matches the implied annual contract size for ORCL, and real margins are obviously not 100%. Payment timing may not be evenly distributed and could be back-loaded, but ORCL depends on OpenAI staying in high growth and maintaining financing access.
Technically, everything above 127 is full of trapped supply. Without a strong OpenAI-side catalyst, the stock is unlikely to repair itself cleanly. It can bounce, but visibility is low.
After a 20% semi drawdown, history supports a rebound
SOXX fell from 655 to 498, a drawdown of more than 20%. Since 2011, the Philadelphia Semiconductor Index has seen 15 drawdowns of more than 20%. Five of those exceeded 30%, and the deepest reached 44%.
JPMorgan's historical data gives the bulls a useful reference. After a drawdown reaches or exceeds 20%, the average return 20 days later is about +8%, and the average return 60 days later is about +11%. That suggests intermediate-term rebound odds are not bad after this kind of washout.
But statistics help with probability, not path. Semiconductor earnings-revision breadth has already turned lower from elevated levels, meaning analyst expectations are starting to come down. More importantly, semis are contributing a growing share of S&P 500 EPS growth. The 2026 Q2 estimate is close to 48%, almost half of total profit growth.
That is the risk. If semi earnings estimates keep falling, the impact will not stay inside SOXX. It will feed into S&P 500 earnings expectations. Semis are one of the market's largest furnaces, now accounting for more than 20% of global equity trading volume. The hotter the furnace burns, the colder the market feels if it goes out.
The hardest zone for SOXX is 550 to 584
Today's 5% bounce in SOXX pushed it right into the 550 to 584 resistance zone. That is not a random area. The prior selloff trapped a lot of capital. Once price rebounds toward cost, many former bulls stop being committed holders and become sellers looking for an exit.
That makes this the cleanest range for shorts. Short sellers using this area usually place stops above 584, while more aggressive traders may use 615. For longs who entered on the recent bottom signal, the practical move after today's bounce is to sell at least half, or most, and lock in gains.
If price breaks 584, the next upside zone is 615. Only after SOXX clears 615 does the next long swing become cleaner. Until then, this is a rebound into resistance, not a confirmed reversal.
Discipline matters more than opinion here. Chasing the maximum gain means accepting the maximum path risk. Waiting for confirmation means giving up some upside in exchange for a cleaner setup.
CTA sell levels are getting closer, so the market cannot only watch semis
The S&P 500 CTA strategy has done very little recently because the index has been moving sideways. In a sideways market, it buys a little. In an up market, it buys a little. When the market falls, it starts selling.
The first sell level is around 7440, possibly adjusted slightly higher toward 7445 after today's move. The second level is 7175, and the third is 6445. The current index is only about 60 points away from the first trigger.
If GOOG disappoints, or if the capex language disappoints semis, the broader market can easily trip that first CTA sell level. That is the real setup today: semis bounced hard, but the broader structure is not comfortable yet. Breadth divergence remains, the big tech and software seesaw remains, and ORCL is showing the credit risk behind the AI infrastructure chain. GOOG capex is not just one company's earnings detail. It is the switch for whether capital still believes the AI trade.