Yields Near 5%, Software Breaks First
US Stocks · Options · News · Views
Oil and Treasuries rose together. SPX lost the first line
The four major indexes fell together on September 10 in New York. Nasdaq and Russell led. SPX closed near 7592, down about 0.6%. It broke 7620, bounced into that line, and got rejected. That is the first CTA sell line. Losing only the first line is not panic. Nine of eleven sectors finished lower. Consumer names and communication services ticked up. VIX tagged about 18 intraday. What had been strong stopped being strong. What was already weak stayed weak. This was not one stock blowing up. Crude and Treasury yields lifted together.
The 10-year jumped to about 4.96%, one step from 5%. The 30-year printed about 5.37%, a roughly 20-year high. Yesterday's $6 billion long-end Treasury buyback did not pin yields. Today they ran harder. In the Middle East, Houthi forces took the Yemeni Red Sea port of Mokha and kept pushing toward the Bab el-Mandeb. Brent reclaimed $100. The gain was about 4% to 6% depending on the print. The strait was not declared closed. Shipping risk is already in the oil price. Oil and bonds up together rewrite growth-stock discount rates first. Tech cannot carry the index alone.
Initial jobless claims were 206,000 for the week. The prior print was revised to about 207,000. The estimate was 205,000. Continuing claims were about 1.774 million, a touch under the estimate. Headline PPI rose 0.4% month over month, in line, and 5.4% year over year versus 5.3%. Core rose 0.2% month over month versus 0.3%, and 4.6% year over year, as expected, faster than the prior reading. More than three-quarters of the goods increase can be booked to energy: diesel, gasoline, jet fuel, heating oil. Core is not running away. Energy is already passing through from the producer side. After PPI, the odds of a 25 basis point hike in September moved to about 70%. At least one hike in the September-October window is almost fully priced. CPI is tomorrow. The FOMC is next Tuesday. Until then, watch the 10-year before the index level.
A 5% to 10% index pullback toward 7000 would still sit inside a reasonable forward-earnings band. It is early. Do not spread the full drawdown map yet. The first line broke. Cut risk. That is not a confirmed trend reversal.
Oracle cleared the numbers. The story did not accelerate
ORCL fiscal 2027 first quarter: revenue $19.3 billion, up 30%. Non-GAAP EPS $1.92 versus about $1.75 expected. Cloud infrastructure $7.4 billion, up 121%, a bit above about $7.2 billion expected. Cloud applications $4.2 billion, up 10%. Software $5.55 billion, down 3%. Hardware about $0.77 billion, up 15%. Services $1.4 billion, up 5%. Non-GAAP operating margin 42%, flat with last year. Remaining performance obligations $664 billion. The company booked more than $30 billion of new AI cloud contracts in the quarter. Operating cash flow was a record about $23 billion. Free cash flow was still negative $5 billion, better than the roughly negative $9.6 billion the Street had. CapEx stayed heavy, about $28.5 billion versus $22.3 billion expected. Oracle also sold about $20 billion of stock through an ATM. Cash is going into data centers. Profit has not caught up. The funding mix is still equity issuance and debt. That structure did not change.
Full-year revenue guidance is at least $90 billion. Non-GAAP EPS is $8.10. Second-quarter EPS guidance is $1.85 to $1.93 in dollars, stripping last year's one-time investment gain. Annualizing this quarter's $7.4 billion of cloud infrastructure is about $30 billion. Getting onto the steeper curve the company has shown before still requires acceleration in the next three quarters. The 2028 and 2030 charts are steeper still. That is a scenario where demand holds and contracts deliver on time. This print cannot prove that. The direction is intact. The pace is not a surprise. Versus MSFT, AMZN, and GOOG, the base is smaller and the growth rate is not closing the gap in one bite. The stock fell about 5% in the regular session and rose about 6% after hours. The day session was recouped. It is still a sideways tape. Support is 147 to 159. Overhead is 175 to 183. Spot is about 153, sitting on the lower band. That clears a "could it be worse" hurdle. It is not a new right-side break. AI monetization looks normal, not exceptional. The turning point is still years out. The next test is whether OpenAI contracts convert on schedule. Contracts and capacity are not the same as revenue following the slope now. A large positive from that side helps Oracle most. A hole there rewrites Oracle first.
Adobe beat. The detail was thin
ADBE fiscal third quarter: revenue $6.76 billion, up 13%. Non-GAAP EPS $6.13, a bit above about $6.08 expected. GAAP operating margin 34.8% versus 36.3% a year ago. Non-GAAP 44.0% versus 46.3%. Creative and productivity monthly active users cleared 1 billion. AI-related annual recurring revenue rose 150% year over year. No dollar figure. The finer the disclosure, the more it reads like confidence. The coarser the number, the more it reads like the pivot is not seated. Fourth-quarter revenue guidance is $6.80 to $6.85 billion. The midpoint sits a touch under the about $6.84 to $6.85 billion Street number. Full-year revenue was nudged to about $26.58 to $26.63 billion. The raise is small. After hours the stock slipped about another 1% to 2%. The forward valuation band barely moved: about 486 to 729 for 2026, about 548 to 822 for November 2027. It looks cheap. The note next to it is value trap. Image and media tools sit on the layer generative models replace most easily. Inside software, that path is the easiest to understand. That is why the multiple stays cheap and the stock does not go. If you are not in, there is no need to force it while visibility is this low. If you are in, you live quarter to quarter on whether recurring revenue holds. A successful pivot leaves room for mean reversion. A value trap means growth keeps slowing and you eat the cost. On the chart, 235 and 248 are the nearby band. The close was about 249. Lose it and the next strong support is 182. The gap in the middle is wide. Do not treat that as a normal dip.
SOXX failed at the door. IGV broke on the open
Yesterday SOXX closed 532, one session short of confirming the falling trendline. Today the high was about 523 and the close about 517. There was not even a retest. Not a near miss. A rejection. Confirmation failed. Trade the group as a hotspot. The trend line is dead until the next real closing confirmation. AMD fell about 3%, closing near 504, on about 15.9 million shares versus about 22.2 million the prior day. The volume break of 498 still holds as a pause, not a washout, unless 467 goes. This is a failed breakout cooling off, not a new downtrend. NVDA closed about 218, low about 217, glued to the lower edge of the 217 to 228 minor support. Minor support is optional. A serious long-term look still waits for 205 to 215.
IGV is cleaner. Support is judged on the open or the close. It opened about 100.8. The gap already broke 101. The close near 101.2 is a bounce after the break. 101 flipped from a buy zone to resistance. Same line, different trend. Before the break, dips there attracted bids. After the break, rallies there attract supply. Treat upside as a rebound. 101 to 107 can be probed. Overhead supply will thicken. Lower lows make a round trip harder. Energy returns only on a break of 107. This software run stopped at 110, about $7 short of a new high. Same level, not the same trend. September and October are already high-vol months. The last two sessions changed character overnight. Short-term work is now intraday. Hold an extra day and the tape can flip. If the map is unclear, sit out.
CPI tomorrow. Flatten the book first
The two after-close prints did not kick the index lower. They also did not lift it back. ORCL lacked surprise. ADBE lacked detail. The 10-year and oil did the repricing. Hedge funds are still in low-exposure watch mode, adding neither longs nor shorts. They are waiting for two things: a clean trend to buy, or a flush that clears risk so leverage can go back on. Spinning with them here has no edge. Sitting out is better than catching a falling knife at 101 and 532, two lines that already failed. Tomorrow's CPI either pulls hike odds back or it does not. If it does not, 5% and the area under 7620 keep chewing. If it does, then you can talk about a second confirmation attempt in semis. Near-term switches: SPX 7620, 10-year 5%, SOXX already failed, IGV 101 is no longer a buy, AMD 467, ORCL 147 to 159.