The call: Neutral-to-bullish into next week — buy AI-infrastructure dips, don't chase the index into PCE + payrolls. Confidence: medium.
Why now: A 2.3% oil drop and a pause at 5.16% on the 10-year let tech carry the market to weekly gains despite 19-year-high yields.
The disagreement: Consensus says 5%+ yields must break stocks — today said strong earnings and 5.1% GDPNow growth can carry them, at least while oil falls.
The level that changes everything: 10-year 5.25% — a daily close above it flips this back to sell-the-rip.
Wall Street ended a scary week for bonds with a relief rally — all three indexes rose and all three finished the week green. Oil falling on Iran-talk hope did more for stocks than any Fed speaker.
Stance: cautiously bullish short-term, neutral medium-term — the market proved it can live with 5% yields for a week; it has not proved it can live with them for a quarter. Confidence: medium.
Indexes (Friday close): The Dow did the heavy lifting into the close.
| Index | Close | Day | Week |
|---|---|---|---|
| S&P 500 | 7,743.41 | +0.51% | +1.2% |
| Dow Jones | 51,828.62 | +478.64 (+0.93%) | +0.3% — snaps 3-week skid |
| Nasdaq Composite | 27,068.72 | +0.5% | +2.1% |
| Russell 2000 (IWM $281.97) | — | +0.11% | lagged — rate pain |
Year-to-date context matters: Nasdaq +16%, S&P 500 +13%, Dow +8% — this is still a bull market, now in its fourth year.
Sectors: Information Technology was the day's and week's hero — up ~0.8% Friday and +3.1% on the week, best of all 11 S&P sectors. Energy was the worst Friday (~-0.9%) as crude fell; rate-sensitive utilities and real estate lagged all week. Only 3 of 11 sectors were green Friday morning — tech did all the work.
VIX: The fear gauge fell to ~15.2, down ~3-4% on the day. In plain English: investors are nervous about bonds but not panicking about stocks. Remarkably calm with the 10-year at a 19-year high — complacency, or confidence in earnings. Watch it.
Treasury yields — the real story: The 10-year touched 5.23% intraday, highest since June 2007, before easing to ~5.16%. The 30-year hit 5.49%, highest since 2004; the 2-year sat near 4.85-4.90%. Translation: mortgages (~7.45% 30-yr fixed, highest since April 2024), corporate borrowing, and government debt all just got more expensive. Stocks rallied because yields paused, not because they fell.
One level for regular investors: S&P 500 ~7,700. We closed just above it (7,743). Hold it through next week's data and the uptrend survives; lose it with the 10-year above 5.25% and the September wobble restarts.
The catalyst was geopolitics, not the Fed. Reports that U.S. and Iranian negotiators discussed a phased deal — reopening the Strait of Hormuz in exchange for easing the U.S. naval blockade, including a reported 7-day plan — knocked WTI down 2.3% to ~$92.41 and Brent down 2.1% to ~$104.32. Lower oil = lower near-term inflation fear = bond-market breathing room.
This strengthened the "good growth can handle high rates" narrative and challenged the "5% yields kill everything" narrative. Evidence for growth: August durable goods flat overall but core capital-goods orders +1.6% m/m and +14.9% y/y, computers +17% y/y; September composite PMI 58.4, a 5-year high; Atlanta Fed GDPNow at 5.1%. The bond market is rising partly for good reasons.
What most investors overlook: the speed of the move. The 10-year rose from below 4.6% in August to 5.23% today — one of the fastest spikes in decades. History says it's the speed, not the level, that breaks things (22V Research: "something always breaks"). Add weak Treasury auctions (soft 5-year and 7-year demand, weak foreign bidding) and Fed officials openly guiding to more hikes (64-66% odds of an October hike) — Friday's calm is fragile.
Real-world link: $4.48 average gasoline (up from $3.16 a year ago), record diesel prices, and 7.45% mortgages are already squeezing households — Michigan sentiment fell to 48.1, down 7% in September, with 1-year inflation expectations jumping to 4.6%. If oil re-spikes, the stock-bond double selloff returns.
Three winners:
1. Microsoft (MSFT, $516.17, +3.7%) — the Dow's engine. A full Copilot overhaul — merging consumer and enterprise apps, adding coding and agent tools with spend controls — plus an Oppenheimer target lift to $570 and an 8% dividend hike. Added ~120 Dow points alone. This is enterprise AI monetization, not chatbot hype.
2. Costco (COST, $922.77, +2.9%) — the beat that needed a second look. Q4 EPS $6.60 vs $6.53 expected, revenue $95.72B vs $94.86B, sales +11.2%. The catch: ~$0.15/share came from a one-time $184M tariff refund, and membership-fee growth slowed ($1.85B, just light). Market forgave it — staples strength on a risk day.
3. Akamai (AKAM, $113.94, +3.2% close; +21% premarket, high $128.46) — the AI-infrastructure surprise. A 7-year, $11.6B deal with Anthropic (expandable to $20B) plus a warrant for ~5% of Akamai at $111.33. JPMorgan lifted to $167 (neutral), Oppenheimer held $180 outperform. Signal: frontier AI labs will pay anyone with ready power + distributed cloud.
Also green: Genius Sports (GENI, $6.44, +11%) on a JPMorgan overweight initiation; People Inc (PPLI, +10-11%) on a reported MGM buyout discussion.
Three losers:
1. Twilio (TWLO, $275.80, -8.0%) — today's sharpest Big Tech-adjacent fall. HSBC cut to Reduce (sell equivalent), $211 target — "expensive vs. better-quality Microsoft/Salesforce" after its rally. Classic high-multiple de-rating on a 5%-yield day.
2. Meta (META, $751.66, -3.3%) — profit-taking after euphoria. Down Friday after +4.5% Thursday; still +~13-17% on the week and knocking on $2T market cap on Muse AI-agent excitement (+27% in three weeks). Up-week, down-day — healthy, not broken.
3. Nike (NKE, $35.75, -0.7%) + MGM Resorts (MGM, $32.58, -3.3%). Nike: Bank of America cut to Underperform with a Street-low $30 target (from $47) — turnaround pushed to fiscal 2028, stock already -40% YTD. MGM: hangover from People Inc walking away from its take-private proposal, compounded by reports MGM may instead bid for People. Salesforce was notably the only Dow stock down >1% (-1.3%).
Most surprising mover: Akamai. A sleepy CDN company just signed one of the largest AI compute contracts of the year. The broader trend: the AI trade is rotating from GPUs to power, pipes, and distributed capacity — Oracle's "force majeure" on its New Mexico data center (stock -1.7% to $137.10) is the mirror image. Centralized mega-builds face permits and debt costs; distributed ready-capacity gets paid now.
1. Actionable for tomorrow/Monday: don't chase Friday's rip — set buy orders under the market. With PCE (Sept 30), JOLTS (Sept 29) and payrolls (Oct 2) ahead plus a 64%+ chance of an October hike, Friday's close near the highs is a poor entry. For short-term traders: fade S&P rips toward 7,760-7,780 if the 10-year holds above 5.10%, targeting 7,620; invalidate on a 10-year daily close below 5.05%. Plain English: let the data come to you.
2. Contrarian: add energy on this oil dip. Everyone sold XLE (-0.9% to $62.04) on Hormuz hope — but Brent still holds $104, Saudi exports just hit 6M bpd (most since the war began), and no deal is signed. If talks fail, oil snaps back fast. For 1-3 month holders: XLE $60-66 is the add zone, $72 target; kill it if Brent closes below $88 (deal actually done). Consensus is short oil into hope; hope has disappointed before.
3. Defensive: own short-duration, not long bonds. With the 10-year at 19-year highs and the 30-year at 5.49%, reaching for duration is catching a falling knife. Park in T-bills / short-term Treasuries and TIPS until the 10-year proves it peaked (two daily closes back below 5.05%). For long-term investors: this protects purchasing power and keeps dry powder for stocks if yields spike again. Avoid passive loans/BDCs — floating-rate credit cracks when "higher for longer" meets an energy shock.
Most important event: August PCE inflation, Tuesday Sept 30 — plus September payrolls Friday Oct 2. Hot PCE + strong payrolls = October hike nearly locked, 10-year toward 5.3-5.4%, stocks retest 7,600. Cool prints = relief rally. JOLTS (Sept 29) and ISM Manufacturing (Oct 1) are the undercard.
One price level: 10-year 5.25%. Thursday's 5.225% and Friday's 5.23% intraday highs both held. A daily close above 5.25% likely forces systematic selling in tech and small-caps; holding below keeps the "orderly rise" story alive.
Three to watch: Micron (earnings Sept 30) — memory prices + the Akamai $1.7B memory pre-buy signal; a guide-up confirms AI hardware still has legs. Nike (earnings Oct 1) — $30 BofA target vs $45 street average; any stabilization is a value-trade spark, another cut confirms consumer stress. Tech (XLK $196.27) vs Energy (XLE $62.04) — the market's tug-of-war in two ETFs; whichever breaks its weekly range first tells you whether this is an AI rally or an oil shock.
The market is mispricing who gets paid for AI: the scarcity is not GPUs, it's permitted power and ready distributed capacity — and Friday repriced only the first name. Akamai's $11.6B (to $20B) Anthropic contract, Oracle declaring force majeure on a flagship data center, and Microsoft pivoting Copilot to enterprise spend-management all point the same way: centralized mega-campuses are hitting financing, permitting, and community walls at 5%+ yields, while anyone with electrons and fiber today can name their price.
That makes distributed-compute and grid-adjacent names — not another GPU proxy — the forward trade into Q3 earnings, with Akamai's Q3 report (early November, watch for $150-300M of 2027 contracted revenue converting) as the proof point. If I'm wrong, Oracle-style delays will spread to distributed players too and the 10-year above 5.25% will compress every AI multiple indiscriminately. Until then, buy the pipes, fade the promises.
Opinion. Timeframes: trades above are 1-3 days to 1-3 months as labeled; no new long-term core change today.
A. Data snapshot · B. Models & assumptions · C. What would prove me wrong · D. Sources
| Item | Level |
|---|---|
| S&P / Dow / Nasdaq | 7,743.41 +0.51% / 51,828.62 +0.93% / 27,068.72 +0.5% |
| ETFs: SPY/QQQ/DIA/IWM | 771.35 +0.54% / 744.50 +0.46% / 517.49 +0.94% / 281.97 +0.11% |
| 10Y / 30Y / 2Y | ~5.16% (high 5.23%) / 5.488% / ~4.85-4.90% |
| VIX / Gold / Bitcoin / Dollar idx | ~15.2 (-3%) / ~$4,315 (+0.5%) / ~$84-85k flat / 101.04 (-0.2%) |
| WTI / Brent | $92.41-92.75 (-2.3%) / $104.32-104.45 (-2.1%) |
| Stocks | MSFT $516.17 +3.7%; COST $922.77 +2.9%; AKAM $113.94 +3.2% (high $128.46); GENI $6.44 +11%; TWLO $275.80 -8.0%; META $751.66 -3.3%; MGM $32.58 -3.3%; NKE $35.75 -0.7%; ORCL $137.10 -1.7%; XLK $196.27 +0.8%; XLE $62.04 -0.9% |
| Macro | Durables flat, core cap +1.6% m/m; Mich sent 48.1 (-7%), 1y infl 4.6%; GDPNow 5.1%; Oct hike odds ~64-66%; gas $4.48; 30-yr mtg 7.45% |
No formal model today — levels-based read. Load-bearing assumption: Friday's yield pause reflects oil-driven inflation relief + orderly positioning, not a top in yields. If wrong (yields resume vertical rise on strong PCE/payrolls), equity support at SPX 7,700 fails and rate-sensitive/growth multiples compress 5-10%.
1) 10-year daily close below 5.05% with PCE cool — bearish-yield thesis breaks, chase risk. 2) Brent daily close below $88 on a signed Hormuz deal — long-energy contrarian breaks. 3) S&P daily close below 7,600 with VIX above 20 — cautious-bull breaks, go defensive. 4) Akamai Q3 shows <$150M of Anthropic revenue converting to 2027 capex — pipes thesis breaks.
CNBC live blog — close, yields, oil, movers
Schwab market update — open, yields, durable goods, week ahead
Investopedia recap — weekly gains, Microsoft, Akamai, Costco, Nike
Edward Jones daily snapshot — yields history, business investment
CNBC — Akamai-Anthropic $11.6B terms
CNBC Investing Club — Microsoft Copilot overhaul
CNBC — midday movers (Genius, Twilio, People)
Costco Q4 2026 results