Cash Is Already Thin; Broadcom CDS Makes a Record as the Tape Stays Lazy
US Stocks · Options · News · Views
The Tape Flipped; SOX Was Weak Again
Friday, August 21, was a broad up day. The Dow gained 0.98% to 53,277. SPX rose 0.43% to 7,674. Nasdaq rose 0.43%. The Dow and the Russell led.
Inside SPX, 334 stocks rose and 167 fell. Equal-weight gained 0.63%. Only two of eleven sectors finished lower, the mirror image of Thursday. SOXX went the other way again and was the weakest sub-group. Semis and software have decoupled this week. SOX looks more like it is fading against the non-semi tape than trading as one AI bloc.
The data split. August S&P Global manufacturing PMI slipped to 53.2 versus 53.9 expected. Services rose from 54.6 to 56.8. The composite rose from 54.5 to 56. Manufacturing slowed on destocking and supply noise. Services lifted the total. A 90-day window to import up to 300,000 tons of beef did not price the session.
Friday reversed Thursday's damage. For the week, SPX still gave back about 1.4%. The index can rise while semis lag. Those are no longer the same line.
Cash Is Spent; SPX Has Not Reached the Risk Levels
The tape has been slow. It does not run far up. It does not wash out far down. Active bids and active offers are both thin. That maps to positioning.
In the latest Bank of America global fund-manager survey, cash is 3.5% of AUM, the sixth-lowest since 1998 and already below the 4% cash-rule line. That line is not an automatic sell. It means cash is so thin that when risk shows up, the main action left is to sell. Optimism is near 1987 readings. ICI mutual-fund cash is also back near historic lows. Hedge funds still have leverage room, long or short. That is risk appetite, not dry powder.
Low cash means the official community is mostly invested. If nobody sells, the tape chops or grinds with seasonal ETF inflows. Volatility stays slow. If risk appears, many will run. Books are heavy, SPX already printed a high, and this week's pullback did almost nothing. Mutual-fund cash cannot refill the bid. Hedge-fund leverage is thrust if the direction is right and an accelerator if it is wrong.
Watch two things. First, VIX. It is the downside-protection gauge built with real premium. 18 to 20 is the warning band. Above 22, downside odds rise sharply. Friday was back near 15. Nobody is paying for protection. Cheap protection is not the same as no risk. It only means no one will write the check yet. Second, the SPX pullback map. The week was down about 1.4% and never tagged 756. Risk starts to rise on a break of 745. It erupts on a break of 729. Spot is still near 7,674. Those are the working retrace levels on this map, not a claim that the index is already at 700. The price is mid-air. There is no must-sell level, so VIX can stay asleep. Complacency grows in that pocket. Stop guessing about a crash every morning. Write 18, 22, 745, and 729 as conditions.
Heavy books mean little cash to push higher. An intact structure means little reason to dump. The tape stays lazy. Either hedge funds lever long and drag it up, or the market waits on next week's events: software prints, NVDA after the close on August 26, Jackson Hole August 27 to 29, and Warsh's first speech there as chair on the 28th. Slide to the risk levels before buying protection. Rest until then. Next week can change the story. It cannot change the fact that this pullback has not touched the gate.
Three branches. If VIX stays under 18 and SPX holds above 756, a grind or a range can persist. Do not treat Friday's bounce as a new trend. The middle path is to wait for Nvidia and Jackson Hole and treat the range as digestion. If VIX tags 18 to 20 then 22 while 745 fails, fully invested books can only sell. That is when the move accelerates. Invalidation is hard: do not call this week's dip a storm before 745 breaks.
AVGO Finances the Sale; CDS Prints a Record
AVGO has had a string of financing headlines. Bloomberg and CNBC have Broadcom seeking about $70 billion to $80 billion in the debt market for AI-chip financing, including buyers such as Anthropic. The wider sketch is a senior piece of about $60 billion to $70 billion, a junior piece of about $30 billion, and a headline some sources stretch to $100 billion. The money may sit in an SPV. Blackstone and Apollo are in the talks. Broadcom may guarantee part of the senior loan. If the counterparty cannot pay, the guarantor pays. That is the same family as the roughly $35 billion June deal, where Broadcom also backstopped part of the senior debt: the chip vendor puts its own credit on the line to move the product.
After the headlines, Broadcom five-year CDS jumped to a record. Friday showed 113.81. Other prints near 122 also circulated. Default insurance now ranks Broadcom fourth behind Chevron at 798, Oracle at 219, and SPCX at 172. Then META 93, NVDA 83, AMZN 61, Google 57, Microsoft 47. A deal that is not even closed already repriced credit protection. The firmer range is $60 billion to $80 billion. $100 billion is still a rumor. Fiscal 2025 revenue was $63.89 billion. If the new year is being talked up toward $100 billion while the firm guarantees debt of that order, the balance sheet is tight. Keep the loan off the parent's books and the risk still comes home to the guarantor.
More chip names are either funding the buyer or guaranteeing the loan to make the sale. Nobody has a clean total for vendor finance. The number is large. The iron is already in the ground. Revenue has to catch up. This wave has to work. Credit spreads will price that before the equity story does. Public debt is already through $40 trillion. If this chain breaks, the damage will not be light. In Q2, Amazon, Google, and Microsoft still showed AI monetization on track. META lagged. That at least moved the Q2 checkpoint out. The next full reset on vendor finance and monetization waits until after the midterms and Q4 earnings. It is early. CDS already shouted.
The equity map is unchanged if high growth holds: 324 to 365 is the zone. Friday closed near 364, up about 1.2%, right at the top of that band. If growth slips and default risk keeps rising, do not hold a break of 324. Use Oracle's tape as the warning. Institutions do not hold the line for you. Buying Broadcom is either a growth ticket or a signature on the buyer's leverage. Know which one.
Tesla Still Needs 368; IGV Tests 107 Again
TSLA traded heavy volume Friday, closed 362.86, high 366.50, up 5.14%. It is one print from ending the upper side of the range. This swing did not tag 270. It stopped near 297. After losing 315 it dropped about 20 points and stalled. A close or open through 368 moves the range up to 315 to 465, a $150 box. Overhead supply is still 415 to 465. A miss by a dollar is still a miss. Until 368, this is still a swing inside the lower box.
IGV repaired the daily RSI bearish divergence that was in place before August 13 by falling and chopping. Friday showed a small pickup in volume, as if money wants another run at resistance. The test is 107, a touch below the prior 108. This is about the third try in the area. A clean break helps software passive flows. A fail is still a range. Component locations differ. An ETF break does not lift every name.
SOXX has also slowed, near 520, still 489 to 532. Versus the broad tape, semis are the cheaper seat. SOX has faded against the index several times this week. A software push from 107 does not have to feed semis. Next is either a test of 532 or a sloppy trip toward the 200-day. Wait for Nvidia next week. NVDA is still 206 to 216, closed 214.72, down about 0.9% for a few sessions. The box has not broken. Treat it as a range. Do not redraw the map every day into the print. Losing 206 is what turns the range into a break.
Friday flipped Thursday. SOX was still the weak line. Institutional cash is already 3.5%. The SPX structure is still intact. Whether this stays boring or accelerates later is a 745 and VIX question, not a Friday up-day question.