Mag 7 YTD analysis and deep dive

Sidekick AI chat created on TrendSpider.

Narrate and compare YTD performance for the Mag 7. Why is Tesla underperforming so heavily, what are the common themes for the leaders, and focus on Meta's recent resurgence. What exactly is Muse and how could it transform Meta's business? Produce an in-depth executive report, weaving in recent transcripts, Segments & KPIs, key fundamentals, news, alternative data, etc.

I'll pull YTD price action myself and dispatch deep research across all seven names in parallel.

Deep research delegation failed (tool error). Since this is a headless sub-chat I cannot re-launch deep research, so I will gather the data directly.

Let me process the data

Let me process the data

4 tools called

YTD % = (close 25 Sep 2026 − close 31 Dec 2025) / close 31 Dec 2025. Example META: (749.005 − 660.09)/660.09 = +13.47%. 3-month base = 25 Jun 2026 close; 1-month base = 25 Aug 2026 close.

Magnificent 7 — 2026 YTD Executive Report

Data as of 25 Sep 2026, 11:46 ET. All prices/fundamentals from TrendSpider data tools.

1. The YTD scoreboard

Ticker YTD 2026 Last 3 mo Last 1 mo Mkt cap
AAPL +24.5% +23.0% +9.2% $4.92T
NVDA +19.9% +14.3% +5.0% $5.43T
META +13.5% +38.0% +31.4% $1.90T
GOOGL +10.0% +0.2% −0.8% $4.13T
AMZN +7.9% +9.7% −4.7% $2.69T
MSFT +6.2% +45.6% +4.5% $3.72T
TSLA −17.5% −1.1% +5.9% $1.50T
SPY (benchmark) +12.7% +4.4% +0.4% —

Equal-weighted Mag 7 = +9.2% YTD, i.e. the group as a basket has underperformed SPY (+12.7%). Only three of seven beat the index. This is the first thing to internalise: "Mag 7 leadership" is not what 2026 actually looks like. Dispersion is enormous — 42 percentage points between AAPL and TSLA.

Note the path, not just the endpoint. MSFT is +6.2% YTD but +45.6% in three months: it collapsed from 481.63 (28 Jan) to 352.83 (25 Jun), −26.7%, then gapped from 390.54 to 451.10 on 30 Jul after the Q4'26 print and has not looked back. META was −20.4% YTD at its 27 Mar low of 525.72 and has since added +42.5%. Anyone telling a clean "AI winners keep winning" story for 2026 is pattern-matching to 2023–25 and ignoring a violent mid-year regime change.


2. Why Tesla is underperforming so heavily

The popular explanations (demand collapse, brand damage, EV competition) are largely wrong on the numbers. Deliveries are fine. The problem is that Tesla has stopped being a profitable company while its multiple still prices it as a hyper-growth one.

Volumes are growing, not shrinking:

Segment Q2'26 Q1'26 Q4'25 Q3'25 Q2'25
Total deliveries 480,126 358,023 418,227 497,099 384,122
Total automotive rev 20.5B 16.2B 17.7B 21.2B 16.7B
Energy gen & storage rev 3.1B 2.4B 3.8B 3.4B 2.8B
Services & other rev 4.6B 3.7B 3.4B 3.5B 3.0B

Q2'26 deliveries +25.0% YoY (480,126 vs 384,122) — a record Q2. Revenue $28.2B, +25.5% YoY, and a 11.9% revenue beat. So the "nobody wants Teslas" thesis is not supported.

Profitability is what broke:

Metric Q2'26 Q1'26 Q4'25 Q2'25
Gross margin 16.8% 21.1% 20.1% 17.2%
Operating margin 1.41% 4.20% 5.66% 4.10%
Net margin 3.99% 2.19% 3.44% 5.29%
Diluted EPS 0.32 0.13 0.24 0.33
Free cash flow −$1.09B +1.4B +1.4B +0.15B
Capex −$5.79B −2.49B −2.39B −2.39B
P/E (ttm, at qtr end) 389x 341x 416x 190x

Operating income in Q2'26 was roughly $397M on $28.2B of revenue. Net income of $1.1B was flattered by a +$1.0B mark-to-market gain on the SpaceX holding, offset by −$300M FX and −$100M Bitcoin. Strip the SpaceX gain and the quarter is close to operating breakeven. Q2 EPS of 0.33 missed consensus 0.44 by −25%. ROE has fallen from 10.4% (Q4'24) to 4.67%.

Management's own explanation, from the 22 Jul 2026 call (CFO Vaibhav Taneja):

  • Auto GM ex-credits "declined sequentially from 19.2% to 16.3%"; without prior-quarter one-offs it would be "approximately flat" — i.e. no underlying improvement at all.
  • Energy gross margin fell "from 39.5% to 20.4%", due to a ~$240M warranty true-up, non-repeating $200M+ tariff benefit, and "ASPs for industrial storage are coming down amidst good, growing competition." He guided energy GM to normalise "in the mid to low 20% range" — a permanent reset from ~39%.
  • Interest-rate subvention: "as interest rates have risen this year, the cost of subvention has risen along with them."
  • Regulatory credits collapsed: $146M in Q2'26 vs $439M in Q2'25 (−67%). That is pure high-margin revenue gone.
  • Capex: "more than $25 billion" for 2026, "will grow for the next two or three years," plus debt facilities "to borrow up to $30 billion." FCF "negative for the quarter."

The narrative-vs-numbers gap. Musk's opening remarks listed production starts for Cybercab, Optimus, Semi, Megapack 3, lithium/cathode refining, and solar cells, plus TeraFab. What the P&L shows is that all of this is currently cost, not revenue. Specific credibility items visible in the data and news flow:

  • Robotaxi scale is measured in miles, not vehicles. Ashok Elluswamy: "more than 380,000 miles of unsupervised robotaxis... across six cities," "zero notable incidents." When Wells Fargo pressed that fleet counts are "in the dozens as opposed to hundreds," the answer was that miles, not vehicles, are the metric. That is a defensible engineering argument and also a convenient one. Tesla's Cybercab fleet was 45 vehicles on 1 Sep, per state records.
  • Cybercab launch landed badly. 3 Sep launch; stock fell; NHTSA opened a probe into whether ~1,000 Cybercabs were properly certified, with 21 questions due 30 Sep. WSJ's Higgins called the promised "storm" a "drizzle." Gary Black called it "largely a bust."
  • Optimus is slipping into reality. Musk on the call: "the initial portion of the S-curve will be quite flat and long"; "there is no supply chain"; "hardest product to scale manufacturing that we've ever made." On 25 Sep, The Information reported Optimus hand and supplier problems. On 21 Sep, Tesla's own AI chief conceded viral Optimus combat clips were "essentially remote controlled."
  • Safety/regulatory drag: record 236 driver-assist crashes reported to regulators in July 2026; ~3M China vehicle recall (21 Aug) over emergency door releases; FSD exceeded Brussels speed limits in 55% of tests ahead of an EU vote.
  • China erosion: China revenue $4.7B in Q2'26 vs $6.7B in Q4'25; China share reported at 6.6%. Offset by Europe (EU registrations +65.9% Jan–Aug, share 1.9% from 1.2%).
  • Sell-side is split to an absurd degree: GLJ $24.86 (Sell, reiterated 4 Sep), Goldman Neutral $360, StoneX Buy $475, Stifel Buy $491. A 19x spread between the lowest and highest target is a statement that nobody can value this on earnings.
  • Q3 risk is live. Goldman cut its Q3 delivery estimate to 435,000 (17 Sep) versus Q3'25 actual of 497,099 — that would be −12.5% YoY. Kalshi traders sat near 480,000. Q3 prints 21 Oct with consensus EPS 0.42 on $27.6B.

Blunt version: Tesla is being repriced from a growth-compounder to a capital-intensive industrial conglomerate in the middle of a self-funded, negative-FCF, multi-year capex cycle, where the high-margin revenue lines (reg credits, energy at 39% GM) are structurally deflating and the offsetting revenue (robotaxi, Optimus, Semi) is not yet material. At 389x trailing earnings it needs the optionality to convert on schedule. Note also the specific bias trap in the bull case: Robotaxi mile growth "more than 10% a week" compounds impressively in a slide deck but currently sits on a base of ~380,000 cumulative miles and dozens of vehicles. Growth rates off tiny bases are not evidence of scale.


3. Common themes among the leaders

Theme 1 — 2026 rewarded capex discipline and punished capex without a visible product. This is the single cleanest pattern in the data. Quarterly capex (purchase of fixed assets) and FCF, most recent quarter:

Ticker Capex, latest qtr FCF, latest qtr YTD
AAPL $2.5B +$31.9B +24.5%
NVDA $2.7B +$21.4B +19.9%
META $30.1B +$1.7B +13.5%
MSFT $35.8B +$19.6B +6.2%
GOOGL $44.9B −$5.9B +10.0%
AMZN $54.2B −$8.8B +7.9%

The two YTD leaders are the two companies with negligible capex and enormous FCF — one that sells the picks (NVDA) and one that mostly opted out of the arms race (AAPL). The two largest spenders with negative free cash flow are the two weakest of the non-Tesla names over the last month. META's September resurgence is precisely the market deciding that its $30B/quarter now has a consumer product attached to it (Section 4).

Theme 2 — earnings genuinely delivered; multiples compressed. 2026 has been a de-rating year masked by EPS growth. P/E (ttm) at quarter-end versus a year prior:

Ticker Latest qtr P/E Year-ago qtr P/E
GOOGL 17.7 18.9
AMZN 19.2 33.5
MSFT 20.8 36.5
META 21.2 26.7
NVDA 26.1 49.4
AAPL 32.5 30.5

NVDA grew revenue +105.9% YoY in Q2'FY27 ($96.2B) and its P/E nearly halved. That is a market paying less for each dollar of AI earnings even as the dollars arrive.

Caution on two headline numbers — do not take these at face value. GOOGL Q2'26 "EPS 9.11 vs 2.87 est, +217% surprise" and AMZN Q2'26 "EPS 5.75 vs 1.82, +216%" are not operating results. Net margins of 93.7% (GOOGL) and 31.2% (AMZN) against operating margins of 34.0% and 13.7% respectively mean the bulk of that net income is non-operating/one-off. It also mechanically deflates their trailing P/Es. Anyone screening on "cheap at 17x" for Alphabet is looking at a distorted denominator.

Theme 3 — operating quality is intact at the top. Gross margins at the leaders are flat-to-expanding: AAPL 46.5% → 50.1% over four quarters; NVDA 72.4% → 75.0%; MSFT ~67–69%; META ~81.4%. TSLA at 16.8% is not in the same business and should not be valued as if it were.

Theme 4 — the macro backdrop was hostile, which explains the de-rating. News flow through Aug–Sep 2026 repeatedly cites: 10-year Treasury yields reclaiming 5% (highest since 2007 on 15 Sep), Brent above $100 (9 Sep) and diesel at $6.06, Fed rate-hike odds at 88% (11 Sep), a BOJ hike, and Big Tech bond issuance projected at a record ~$320B this year with bond investors demanding higher yields from GOOGL, META and NVDA. Goldman is quoted (25 Sep) seeing hyperscaler AI capex rising 50% to $1.2 trillion. Multiple credible voices are now openly bearish on the capex cycle: Dalio ("we are now in a bubble"), Gundlach exiting Big Tech, Fidelity's Timmer calling the AI trade "dead money," BofA's September survey showing 42% see AI capex as the likeliest systemic credit event. Note these are attributions by others, not causal claims.

Theme 5 — an AI-safety political overlay arrived in September. 13–14 Sep saw a broad "pace the frontier" episode (Anthropic, OpenAI, SpaceXAI executives calling for slowdown) that hit semis, plus EU child-safety rules, a Trump/Congress AI-oversight meeting set for 29 Sep, and state-level pushback on data-center tax breaks. This is a new, non-fundamental risk factor for every name here.


4. Meta's resurgence — and what Muse actually is

The move. META bottomed at 525.72 on 27 Mar (−20.4% YTD), ground sideways in the 540–600 range through August, and then: 572.34 (31 Aug) → 749.005 (25 Sep) = +30.9% in 18 trading sessions. From the 27 Mar low, +42.5%. On 21 Sep alone it added ~$207.5B in value (+12%). On 24 Sep it passed SpaceX to become the world's seventh most valuable company at ~$1.98T. It printed a new 52-week high on 25 Sep.

Why this is remarkable: it happened after a bad quarter. Q2'26 (reported 29 Jul) was Meta's only EPS miss of the year — 6.18 vs 7.18 est, −13.9% — with operating margin collapsing to 30.9% from 43.0% a year earlier, FCF down to $1.7B from $9.0B, and capex up 82% YoY to $30.1B. The stock fell 585.61 → 539.03 on 30 Jul. The September re-rating is therefore not an earnings-driven move. It is the market re-underwriting the capex.

What Muse is. Muse is Meta's personal AI agent — an agentic assistant, not a chatbot and not a device — announced 8 Sep 2026 as "the world's first personal AI agent built for everyone." It is the consumer-facing top of a whole Muse-branded stack that Meta built out through H2 2026:

Component Type Date
Muse Image image-generation model 7 Jul 2026
Muse Spark 1.1 / 1.2 frontier foundation models 10 Jul / 5 Aug
Muse Code coding agent on Spark 1.2 5 Aug 2026
Muse Glimmer open-weight agentic model, runs locally on Mac/PC 10 Aug 2026
Muse Video, Voice Mode multimodal in development
Muse (personal agent) consumer agent app 8 Sep 2026
Muse Charm referenced by JPMorgan in wearables context 24 Sep 2026

What it does and how it is monetised. Muse executes tasks on the user's behalf — including placing phone calls and, critically, shopping and checking out. Monetisation is emerging along three lines:

  1. Agentic commerce. In four days Meta announced SHOP Shopify (22 Sep), PYPL PayPal checkout across merchants worldwide (22 Sep), Expedia (22 Sep), Instacart (22 Sep), then Walmart, Sephora and Best Buy (24 Sep). Meta is positioning Muse as the transaction layer between 3.6B users and retail.
  2. Subscription. Meta One launched 15 Sep globally at $2.99/mo single product, $7.99 individual bundle, $14.99 creator/business bundle, with core Meta AI staying free. Zuckerberg said on 24 Sep that Muse will remain free "for a huge number of tokens."
  3. Advertising. The existing engine — still 97.7% of revenue (Q2'26 ads $59.4B of $60.8B) — gets a new, higher-intent surface.

Why it matters to the P&L. Meta's revenue engine, Q2'26 vs Q2'25:

Segment Q2'26 Q1'26 Q4'25 Q2'25
Advertising revenue 59.4B 55.0B 58.1B 46.6B
Other (FoA) 1.0B 885M 801M 583M
Reality Labs revenue 431M 402M 955M 370M
FoA operating income 23.4B 26.9B 30.8B 25.0B
Reality Labs op income −4.62B −4.03B −6.02B −4.53B
Daily Active People 3.60B 3.56B 3.58B 3.48B
ARPP $16.86 $15.66 $16.56 $13.65
WW impressions YoY +14% +19% +18% +11%
WW price per ad YoY +12% +12% +6% +9%

Ads grew +27.5% YoY with both volume (+14% impressions) and price (+12%) contributing — the healthiest possible mix. ARPP is +23.5% YoY. Reality Labs remains a ~$4.6B/quarter hole against $431M of revenue: Reality Labs revenue is 0.7% of total while consuming ~$18B/yr of operating income. That is the segment concentration risk in one line, and it is why Muse matters: it offers a monetisation path for the AI/compute spend that does not require the metaverse to work.

What management has actually said — and the honest gap. Meta's most recent earnings call was 29 Jul 2026, six weeks before Muse launched. Muse is therefore not discussed on any earnings call to date, and the transcript I retrieved is only the closing portion of that call. What Zuckerberg did say there frames the bet:

> "having kind of sovereignty over building your own models is going to be an important part of that stack going forward" > "if you're trying to build personal superintelligence for people or you're trying to build a business agent for small businesses, that needs to have potentially some different skills than what the other labs are tuning" > "I get that this is a big investment and it's a big bet. We see the technology working... My personal bet is that the people who invest in this are going to be rewarded and feel very good over time." > "we are today, and expect to be in the foreseeable future, demand constrained... we still have numerous ROI-positive places that we would put compute toward if we had it." > On 2028 planning: "we're really focusing on flexibility. That's giving us the ability to have land and power, but to really make the actual decisions about buying chips and other big-ticket items further in the future."

That last point is a genuine tell: Meta is committing to land and power but deliberately not committing to silicon beyond 2027. That is a hedge, and a sensible one.

The bear case on Muse is documented, not hypothetical. Give it equal weight:

  • 9 Sep: internal tests reportedly revealed privacy risks, security flaws and "failure modes that made it unreliable."
  • 22 Sep (Reuters): Meta is testing a "human concierge" — human contractors quietly handling some phone calls placed via Muse. On 23 Sep, tests showed human callers outperformed Muse alone. An agent propped up by humans is not yet an agent.
  • 22 Sep: Amazon blocked Muse from shopping; two days later it opened seller data to Claude. Chamath: "everyone will start blocking everyone." The single largest US e-commerce surface is off-limits.
  • 5 Aug: Muse Spark 1.1 breached another company's systems during a cybersecurity test.
  • Talent attrition: Jiahui Yu (ex-Gemini/GPT-4o) left 14 Aug; Andrew Tulloch left 9 Sep.
  • Timing of revenue: BofA — Muse beat ChatGPT's early iOS download pace, but "meaningful revenue may not arrive before 2028." Wedbush — monetisation "somewhat limited" near term, subscriptions ~$5B by 2027, and it still rates the stock Hold at $650. Gene Munster noted "weak Muse adoption data" ahead of Connect on 23 Sep and admits he missed the move. Allen Bond: investors "still need proof that Meta's AI spending can generate sufficient returns."

How Muse could transform the business — stated as a mechanism, not a forecast. If Muse becomes the default agent for a meaningful share of 3.6B DAP, Meta converts from an attention business (sell impressions at $16.86 ARPP) to a transaction business (sit in the checkout flow, take commerce economics), plus a subscription line, plus a justification for compute spend it is already making. That is the re-rating thesis the sell-side is expressing. The second-order evidence is telling: on 22–23 Sep, SHOP and PYPL rose on partnership news while GOOGL, INTU and subscription-model names fell on disruption fear, and AMD crossed $1T on an agent-drives-CPU-demand thesis. The market is pricing Muse as a platform shift, not a feature.

Where it is unverified: I cannot confirm actual Muse DAU/MAU, revenue, token costs, unit economics, or how much of "Muse" inference runs on Meta's in-house silicon. Nothing in the fundamentals data yet reflects Muse at all — the first quarter that can is Q3'26, reporting 28 Oct 2026 (consensus EPS 6.38 on $63.3B). That print is the entire thesis test.


5. Earnings record, 2026 YTD

Ticker Latest report EPS act/est Surprise Rev YoY Next report
AAPL 30 Jul 2.02 / 1.89 +6.9% +16.4% 29 Oct
MSFT 29 Jul 4.74 / 4.23 +12.1% +17.8% 28 Oct
GOOGL 22 Jul 9.11 / 2.87 +217%* +24.2% 28 Oct
AMZN 30 Jul 5.75 / 1.82 +216%* +19.6% 29 Oct
NVDA 26 Aug 2.22 / 2.09 +6.2% +105.9% 18 Nov
META 29 Jul 6.18 / 7.18 −13.9% +28.0% 28 Oct
TSLA 22 Jul 0.33 / 0.44 −25.0% +25.5% 21 Oct

*Inflated by non-operating items — see Section 3.

Five of seven report in a 10-day window (21–29 Oct). That is a concentrated event-risk cluster, and for META specifically it is the first Muse datapoint.


6. Alternative data and positioning

Analyst ratings. META has seen a near-unanimous target stampede in 15 days: JPMorgan upgraded Neutral→Overweight on 10 Sep (640→820) and raised again to 920 on 24 Sep; Tigress 995, Canaccord 950, KeyBanc 900 (from 780), Rosenblatt 886, Citizens 885 (from 770), Jefferies 875 (from 710), Piper 875, TD Cowen 865, Cantor 860 (from 680), Raymond James 860 (from 650), Evercore 860, BofA 810, Citi 800, Wells Fargo 796 (from 640), RBC 770. Holdouts: Wedbush Hold $650, StoneX Hold, Needham Hold, Benchmark Hold. Note the pattern — most of these are target revisions, not rating upgrades, and many came after a 12% single-day move. That is momentum-chasing, and it should lower your confidence in targets as independent information.

TSLA ratings since 1 Aug: only five records — Stifel Buy $491 (cut from 508), GLJ Sell $24.86 (twice), StoneX Buy $475, Goldman Hold $360. Sparse coverage activity plus a 19x target spread.

Insider activity. META — the only >$5M transactions since 1 Jun are sells, and they cluster into the rally: Chris Cox (CPO) sold $13.0M (9 Sep), $13.5M (15 Sep) and $28.5M (21 Sep) — 40,000 shares on the day the stock jumped 12% — leaving 184,516. Susan Li (CFO) sold $5.1M on 18 Aug. These are stock-appreciation-right dispositions and look schedule-like in cadence, but the 21 Sep size step-up into strength is worth noting rather than dismissing. TSLA: no insider transactions above $5M since 1 Jun — no open-market buying either.

Hedge funds / 13F (Q2 filings, Jul–Aug 2026). META net buying dominates: JPMorgan +45.9M shares ($25.9B), Invesco +22.8M ($12.8B), Sixth Street +15.4M ($8.7B new), Capital Research +8.7M ($4.9B), BlackRock +4.6M, State Street +3.5M, Pershing Square new 3.2M shares ($1.8B), Renaissance, D.E. Shaw, Dodge & Cox, WCM all adding. Sellers: FMR −20.7M ($11.7B), Wellington −4.5M, T. Rowe −2.2M, Baillie Gifford −1.2M. Ark Invest has been buying publicly through September ($28.1M on 9 Sep) while selling Alphabet. TSLA is mixed-to-positive at the custodian level (JPMorgan +42.0M, Invesco +38.5M, SG Americas +9.5M, BlackRock +6.0M) but the active-manager signal is negative: Viking Global cut 78%, down to 547.8K shares; ExodusPoint effectively liquidated (−4.9M to 20.1K); Jennison, Amundi, Susquehanna trimmed. Distinguish index/custody flows from conviction flows — the latter is leaving Tesla.

Retail participation and sentiment (% of daily volume; sentiment −100..+100). META retail share more than doubled into the rally — 0.83–1.16% mid-September to 2.73% on 21 Sep and 2.55% on 24 Sep — but net retail sentiment stayed negative throughout (−4 to −3). Retail was net selling into the move; institutions drove it. TSLA retail share is steady at 1.4–1.9%, sentiment −1 to −4. AAPL has the most negative retail sentiment in the group (−9 to −7 on 21–24 Sep) despite being the YTD leader. AMZN is the only clearly net-bought name (+1 to +4).

Short volume. META short volume roughly tripled into the breakout: 2.6–3.9M shares/day on 16–17 Sep to 10.6M on 21 Sep and 9.9M on 24 Sep. Some is hedging/market-making against the move, but shorts are stepping in front of it. TSLA short volume is flat-to-lower: 12.2M (18 Sep) down to 7.7M (24 Sep) — consistent with the 23 Sep "short squeeze fuels rally" report and the +5.9% one-month bounce.

Dark pool. Data returned only a single week (w/o 17 Aug): META 45.3M shares, TSLA 91.8M. Insufficient history to establish a trend — I will not infer accumulation or distribution from one datapoint.

Not retrieved: congressional/government trades and CNBC sentiment were not pulled for this report; treat those channels as unexamined.


7. Verdict

What is actually driving 2026 (attributions by management and the sell-side, not causal claims of mine):

  1. Capex scrutiny is the axis of the whole year. Every leader either spends little (AAPL, NVDA) or has newly demonstrated a product attached to the spend (META). Every laggard is spending heavily into negative FCF without a visible consumer payoff (AMZN, and MSFT/GOOGL for most of H1).
  2. Multiple compression, not earnings disappointment. Earnings delivered — six of seven grew revenue 16–106% YoY. P/Es fell hard. Rising yields (10Y >5%), oil >$100, hike odds at 88%, and record Big Tech bond issuance are the cited mechanism.
  3. META: a product event resolved an ROI question. Muse plus an agentic-commerce partner land-grab (Shopify, PayPal, Walmart, Instacart, Expedia, Sephora, Best Buy) reframed $30B/quarter of capex from cost to investment. JPMorgan calls it potentially "AI's biggest consumer app yet."
  4. TSLA: a margin and capex story wearing a robotics costume. Volumes grew 25%; operating margin fell to 1.41%; FCF went negative; energy GM reset from 39.5% to low-20s; reg credits −67%. The optionality has not begun converting.

Biggest risks to the current trajectory:

  • META has priced a lot on ~17 days of app-store data. Retail was net-selling, short volume tripled, the CPO sold $28.5M into the spike, targets were raised after the move, Amazon has blocked Muse, and Meta is reportedly using human contractors to make Muse's calls work. The 28 Oct print is the referee. A Q3 miss or the absence of Muse KPIs would be ugly from a 52-week high.
  • Whole-group AI capex risk. Goldman sees hyperscaler capex +50% to $1.2T; BofA's survey has 42% flagging AI capex as the likeliest systemic credit event; bond investors are already demanding more yield from META, GOOGL and NVDA. A funding-cost shock hits the heavy spenders first.
  • AI-safety regulation became a real factor in September (EU under-15 rules, 29 Sep White House/Congress meeting, state data-center tax fights, Meta's own Oversight Board demanding nine policy changes).
  • TSLA Q3 on 21 Oct. Goldman at 435,000 deliveries vs 497,099 a year ago. First YoY decline plus another negative-FCF quarter against a 389x multiple is a poor combination. NHTSA's Cybercab response is due 30 Sep.
  • Concentration: META is 97.7% advertising; Reality Labs burns ~$18B/yr on 0.7% of revenue. TSLA is ~73% automotive with sub-17% gross margin.

Where the data is ambiguous or missing: (a) The Meta earnings-call transcript retrieved was only the tail of the 29 Jul call, and Muse post-dates it entirely — so there is no management guidance on Muse economics in any filing or call. (b) GOOGL and AMZN Q2 net income and their trailing P/Es are distorted by unidentified non-operating gains; I could not decompose them with the available metrics. (c) Dark pool history returned one week only. (d) Peer-relative valuation benchmarks, relative-strength percentiles, and per-ticker deep dives for AAPL/MSFT/GOOGL/AMZN/NVDA were not obtained — the parallel research pass failed on all seven tickers and this report was rebuilt from direct data pulls, which are thinner on those five names than on META and TSLA. (e) Reported P/E values are as of quarter-end, not today's prices.

This is analysis, not trading advice. I can't make decisions for you, and none of the above should be treated as a recommendation to buy or sell.

Sidekick is a TrendSpider AI agent. It has the entire set of data available on TrendSpider at its disposal. It can do research, design scans for you, create alerts and automate your routine. Sidekick is a TrendSpider AI agent that can do the job for you. Login & Chat