Pre-Market Briefing on Oct 8: PENG Surges 13% on Strong FY Guidance with 30-Year Treasury Auction Eyed

All three major indexes closed lower on Wednesday, October 7: the S&P 500 at 7,801.15 for 0.23%, the Nasdaq at 27,538.69 for 0.22% and the Dow at 51,179.87 for 0.66%. The stock in focus is Penguin Solutions (PENG), which closed at $72.61 for a 13.08% gain, adding about $430M of market value to reach $3.7B on 20.97 million shares, 7.1 times average volume. Fourth-quarter revenue was $567M, up 68%, with adjusted earnings of $1.00 a share against $0.78 expected; what moved the price was the new fiscal year's guidance midpoint of $2.43B, a tenth above the $2.20B modelled. ⚠️ That guidance is a range of $2.25B to $2.6B, $350M wide, and guidance is not realised revenue. Today, October 8, weekly jobless claims land at 12:30 UTC and a $22B 30-year Treasury reopening prices at 17:00 UTC. Data is as of the October 7 US close, and all times are UTC.

Unit: %, single-day change on October 7; all three fell, so the zero line sits on the right. The Dow was furthest down at −0.66%, the S&P 500 −0.23% and the Nasdaq −0.22%.

 

The Dow fell close to three times as far as the other two, but that does not make blue chips weaker than tech. The Dow is price weighted, so one high-priced share takes a large slice of it; the other two are capitalisation weighted, and the same name's effect is diluted.

 

When three indexes separate by that much, ask whether they are weighted the same way. Price and capitalisation weighting price the same event differently.

 

Unit: index points; the lower bar is part of the upper one, not a second item. The Dow lost 341 points, of which Caterpillar (CAT) alone contributed about 295.

 

Caterpillar fell 5.75% to $813.83 after a rating downgrade, and removing it leaves the Dow roughly flat: a rating event at one company, not a sector retreating. Nasdaq-listed technology names averaged 1.14% and utilities 2.26% the same session (sample limited to Nasdaq-listed constituents), while Micron (MU) rose 4.06% to $1,088 and Intel (INTC) 0.55%.

 

When an index falls and the group you follow rises, find the largest contributor first. Losses in one name are a stock event; losses across twenty or thirty names are a sector retreating.

 

Unit: %. The basis is the card's: the five largest companies by market value in PENG's Nasdaq semiconductor industry, plus PENG itself, a hand-picked comparison set. PENG sits at one end on +13.08% and TSMC at the other on −2.09%, three up and three down.

 

Keep the two figures apart. That industry field averaged 1.87%, so PENG ran 11.2 points ahead, but the 1.87% is a field average, not an average of the six on the chart, and both holding at once tells you the field covers far more companies than those six. So PENG neither moved against its group, since the field average is positive, nor led it, since most of the set fell.

 

When an average and a comparison chart disagree, neither is wrong; the denominators differ. Confirm what the average samples before using it as a baseline for that chart.

 

Unit: 0–100, against peers and its own one-year history, over a one-year window rather than the single session. Peer rank and relative strength both read 100, trend position 76.6 and valuation heat 68.8; the weakest axis, volatility control, reads 5.8.

 

Four corners reach the rim and the one gap is deeper than anything else on the chart. Its beta is 2.85, about three times the broad market: the property that let it run 11.2 points ahead in a day is the amplifier on the way down. It still sits 19% below its 52-week high.

 

Separate the axes measuring position within the group from those measuring position in its own year. Strongest today and not standing high can both hold; volatility control describes the path, not the result.

 

The supporting name is Super Micro (SMCI), up 3.41%: the Wall Street Journal reported AI cloud provider Lambda is raising about $4B, backlog up from $15B in June to $50B in September.

 

Two things today: weekly jobless claims at 12:30 UTC, expectation 200,000, before the open, and the 30-year auction result at 17:00 UTC, a $22B reopening, mid-session. No technology earnings today, only consumer names such as PepsiCo (PEP). With no CPI and no payrolls this week, the September minutes out and the October FOMC on the 27th and 28th, the long end is the only pricing variable still moving.

Unit: %, the high clearing yield at each of four consecutive sales from June to September; the bars start at 4.95% rather than zero. The yield climbed from 5.020% to 5.308%, a total of 28.8 basis points.

 

Why read the clearing yield rather than the screen quote: it is the number real money settled at, while a quote is only a quote. The 30-year on screen was already 5.67% on October 7, about 36 basis points above September's clearing yield and in the high range since 2001, and buyers absorb $22B at that level today. A data-centre asset takes 15 to 20 years to pay back, so 36 basis points on the discount rate cuts the present value of the far years far more than the near.

 

For the long end, read the clearing yield that settled, not the screen quote alone. A quote shows willingness; a clearing yield shows someone actually paid.

 

Unit: %, one day across maturities, a different basis from the auction history above, so the two cannot be read across. The 2-year reads 4.77, the 5-year 5.03, the 10-year 5.28, the 20-year 5.71 and the 30-year 5.67, 90 basis points between the 2-year and the 30-year.

 

The curve peaks at the 20-year rather than the 30-year, and that is the cell worth remembering today. Policy rates hold only the short end, the target at 3.75% to 4.00% after a 25 basis point hike on September 16 carried 12 votes to none, while everything past ten years is buyers asking for more compensation. AI capital spending pays back inside roughly 10 to 20 years, the stretch that rose hardest.

 

When the long end rises, find which cell the curve peaks at first. Where the peak sits decides whether the move came down from policy or the market is repricing one band of maturities itself.

 

Two steps today: compare the clearing yield with the 5.67% screen level, where anything above means buyers wanted a discount; then watch whether the bid-to-cover holds its one-year average of about 2.40, which was 2.61 in September and signals weaker demand only near 2.3. The result follows two minutes after bidding closes.

 

These four are placed by hand according to what each business does, not by exchange classification. Mining rigs map onto Broadcom (AVGO) and its custom silicon, graphics cards onto Nvidia (NVDA), rented compute onto CoreWeave (CRWV), and holding coins onto Strategy (MSTR), which issues shares and debt to buy coins.

 

The whole difference is when the money arrives. Nvidia and Broadcom are paid on shipment; CoreWeave borrows first to buy the cards, with second-quarter capex at 2.5 times revenue and net interest expense 2.4 times a year earlier, and collects over multi-year contracts; Strategy has no customer here, holding 848,000 bitcoin at about $75,000 a coin, so the asset price is the share price. Broadcom carries one more layer: 73% of last quarter's AI chip revenue came from accelerators built for six customers, and if one goes, nothing replaces the gap.

 

Inside one story, ask what it sells and to whom, how hard it is to switch away, and when the cash lands. Answer those three and four companies separate into four types.

Unit: %, the 52-week high minus the latest close, divided by the high and multiplied by 100; shares are taken at the October 7 US close and bitcoin at the same date's UTC daily close, two windows about four hours apart, so only one decimal is carried. Nvidia reads 2.4, Broadcom 23.9, bitcoin 34.3, CoreWeave 42.3 and Strategy 54.4.

 

One AI story, and the drawdown stretches from 2.4% to 54.4%. What differs is not the story but those four business structures. The line worth looking at is Strategy sitting 20.1 points deeper than the bitcoin it holds, and that gap comes from equity and premium, not the coin. ⚠️ This is a drawdown rather than a gain, and a higher reading is not a quality rating or a ranked list.

 

When one story contains drawdowns tens of points apart, go back to business structure before valuation. When the cash lands and how firmly customers are held decide how long a story carries a share price.

 

A: Neither. Moving against the group needs a negative field average, and this was positive; leading needs most of the group to rise, and three of the six fell with TSMC down 2.09%. State the number: PENG ran 11.2 points ahead of the field average.

 

A: Those figures explain only part of it. What took the shares up 13% was the new fiscal year's guidance midpoint of $2.43B, a tenth above the $2.20B modelled. On an earnings day, compare the midpoint with the model first.

 

A: The move did not come down from policy rates. Policy holds only the short end, and everything past ten years is buyers asking for more compensation, while AI capital spending pays back inside 10 to 20 years, the stretch repriced most.

 

A: The bid-to-cover. Its one-year average is about 2.40 and September came in at 2.61; near 2.3 would signal weaker demand. The clearing yield is read against the 5.67% on screen, where anything above means buyers wanted a discount.

 

A: No, what it holds is bitcoin. The extra 20.1 points come from the equity and premium layer, not the coin: the same asset in a different business structure moves differently.

 

Disclaimer: This article is compiled and written by the MEXC RealStocks team. The data in this article is based on the closing of the US stock market on October 7, 2026. The content is a compilation of public market information, and individual stocks are publicly discussed targets, which do not represent the recommendation or opinion of MEXC and do not constitute any investment advice. More US stock content: @MEXC | @Alpha_MEXC | @MEXC_Research