# Special Situations Report — Weekly Call (readable notes)

- **Title:** SSR Macro — FOMC decision week with CME odds of a 25bp hike at **87%** (Singh: "I was not expecting a hike, but now it looks like there's a high probability") after a hot **3.4% y/y headline CPI** and in-line core CPI/PPI; the **10-year at 4.96%** and a tailed **30-year auction at 5.308%** as Bessent's **$6B** long-end buyback (only **$5.19B** executed) is rejected by the market; **Red Sea chaos for Saudi** — drones shut the **1,200km East-West pipeline** (4-5 mb/d of Hormuz bypass), Saudi August output at **6.2 mb/d, the lowest since 1990**, Brent **~$108** / WTI **~$103**, diesel above **$6/gal** for the first time ever, the Oman GCC-Iran meeting postponed and tanker rates at a record **$800k/day**; the S&P only ~3% off its high on a **12% Mag 7 rebound** while just **17%** of members sit above their 20-day average; **Dario Amodei's "pace the frontier" letter and Altman's "ill-advised moment to go public"** — and, in the book: **SK Telecom (SKM) exited** at $38-40 with its Anthropic stake fully priced at a $1.7T mark; a new **Valley National (VLY) CRE short** as the rate-hike hedge; **EVC and PRTH DCF models** (~59% upside; $21 fair value); an **Air Canada odd-lot Dutch tender** (C$29-33/34 vs C$27.53); adds to **Redwood Trust (RWT)** equity and baby bonds on a convert-hedging sell-off; **Reservoir Media (RSVR)** bought in the low $9s under a $10.50 insider bid; plus **long IJR / short IWM**, the **Kinetik (KNTK)** sale on watch, and **KO / MCD** as defensive names he is not buying
- **Show:** Weekly SSR research call (premium subscriber recording — no public video)
- **Guest/host:** Jay Singh (founder, Special Situations Report; ex-Goldman Sachs)
- **Date:** 2026-SEP-13 (Sunday call; the deck is dated 9-13-26 and the transcript title garbles the brand as "2026-09-13 - SSN")
- **Source files (this folder):** `transcript.pdf` (the verbatim premium transcript, section headers carrying the recording's `MM:SS` cues), `report.pdf` (the condensed report + actionable items + Q&A write-up), `agenda.pdf` (image-only one-page agenda) and the **54-page weekly deck** `SSR Macro - FOMC Decision, PPI, CPI Meet Exp, 10Y Spikes with Oil, Red Sea Chaos for Saudi, PRTH EVC Models, SKM Anthropic Sale on Slowdown, AC Dutch Tender, RWT Add on Con.pdf` (32-item numbered agenda on its page 1, which is the same list as `agenda.pdf`). Also sent with the call and saved here: the analyst DCF workbooks **`EVC_Quarterly_3_Statement_FCFE_Model_with_DCF Valuation and Sensitivities.xlsx`** and **`PRTH_Quarterly_3_Statement_DCF_Model_with_Segment Financials.xlsx`**; the screening list **`MS Non-AI Top Picks.xlsx`** (Morgan Stanley's global "mostly non-AI" picks, deck p. 14); and third-party research PDFs **BlackRock — *Why Japan Matters to US Bond Investors***, **Deutsche Bank — *AI infrastructure investment and private capital***, **Fidelity — *US Fixed Income, Sept 2026***, **Franklin Templeton — *Mapping China's AI Ecosystem***, **J.P. Morgan — *on the Iran War, AI Transformation and Tracking*** and **Morgan Stanley — *3Q Update on Equities and Bonds***. (He also names a Standard Chartered "inching towards a Fed hike" note and a ~200-page Morgan Stanley report on China's next industrial revolution, posted to the SSR Q&A tab but not in this folder.)
- **Note:** readable notes extracted from the premium PDFs; fillers (um/uh/you know/false starts/stutters) removed, wording otherwise verbatim from the recording, with a few passages lightly condensed where the speaker repeats himself. Section headers carry the recording's `MM:SS` cue; the "Ref" column on the analysis page uses the same plain-text times (plus deck page numbers), because there is **no public video to deep-link into**. **Timestamp caveat:** the transcript's section cues jump from **16:37** straight to **1:03:04** — the whole VLY short, SKM exit, the AI-pacing discussion, RWT, RSVR, the Morgan Stanley list and the Iran war update all sit under the 16:37 header, so the sub-headings below cite deck pages instead of invented times. Auto-transcript garbles corrected and listed here: **"SSN" = SSR**; **"EVBC" = EVC (Entravision)** — the report compounds the garble into "Everyday Health / EVC (EVBC)", which is wrong: the model, the deck (p. 5) and the prior two calls are all Entravision; **"AC.CN" = Air Canada's TSX line (AC / AC.TO)**; **"Adario" / "Dario Amadai" = Dario Amodei**; **"MATR" = METR** (the deck's spelling); **"Kevin Worth" / "Worsh" = Kevin Warsh**; **"Hasset" = Kevin Hassett**; **"Hemmick" = Cleveland Fed President Beth Hammack**; **"Besant" = Scott Bessent**; **"Sane Takashi" = PM Sanae Takaichi**; **"Breen Capital" = Brean Capital**; **"WestBill Richmond Hill" = Wesbild / Richmond Hill** (the deck's spelling; the report's "Westaim Corporation" is unsupported by the transcript and deck); **"Irenic Capital"** as spoken; **"NCUNP" = NSC/UNP**; **"Warner Brothers peace guy" = WBD/PSKY** (the Paramount Skydance deal); **"FOKA" (deck) = FOXA**; **"Kinetic" = Kinetik Holdings (KNTK)**; **"Trip Advisors, or trip.com"** is left as spoken (the deck calendar is not legible on it); **"Lauren" / "Lawrence Alexander" = Jefferies' Laurence Alexander**; **"SLS" = SOLS** (deck); **"Actin Technology… with two Cs" = Accton Technology** (Taiwan); **"Jacob Coxson" = Jacob Coxon** (deck); **"CT1" = CET1**; **"WTQ" / "RWTTS" = the Redwood baby bonds the deck prints as "RWTQ and RWTS"** (the report adds "RWTN"; none is resolved to a verified symbol here, so they are kept inside the RWT row). Internal errors and inconsistencies, left as spoken and flagged: the **Air Canada tender range is "between 29 and 34" on the call and in the report but C$29.00-33.00 in the deck**, and his return math ("33… divided by 23.53… 17%") does not reconcile — C$33 / C$27.53 is **~20%** and C$34 is **~23.5%** (the report's figure); repurchase capacity is **"about 28 million shares, ~10%"** on the call vs **27.59M (~9.8%) of a C$800M substantial issuer bid** in the deck, and the funding is "a recent **sale** of their Aeroplan loyalty platform" on the call vs "monetization of a **minority stake** in the Aeroplan loyalty platform" in the deck; **VLY's non-owner-occupied CRE ratio is given as "325%", "329%" and "333%" on the call, 329% in one deck bullet and 317% (Q2 2026) in the deck's peer table**, with the 2023 peak given as both **474%** and **476%**; **"one of the biggest insider sales for Uber"** describes the COO's **purchase** of 70,000 shares; **OPEC's cut to 2026 demand growth is "38,000 b/d" on the call and in the report but 380,000 b/d in the deck**; **August headline CPI printed 3.4% y/y**, although the prior call's calendar had y/y *falling* to 2.4% — the recording does not reconcile the two; the **Anthropic IPO is "October 26th" on the call vs "October 2026" in the deck**; PRTH's DCF value is self-corrected mid-sentence from "$20 billion" to "about 2.6 billion dollars… about $21 a share"; the **CEO's PRTH stake is "like 60%"** here vs 56.5-58% on the AUG-23/30 calls; **"Oxtay, we talked about"** in the special-situations list resolves to no name and is left unresolved. The transcript footer's "Date of Transcript: 2024-05-24" and its "ssn_transcript_v6 - 2026-01-26" prompt name are stale template artifacts — the real call date is **2026-09-13** (Sunday); the recording runs **~1 h 53 m**.

---

2026-09-13 - SSR - FOMC Decision, PPI, CPI Meet Exp, 10Y Spikes with Oil, Red Sea Chaos for Saudi, PRTH EVC Models, SKM Anthropic Sale on Slowdown, AC Dutch Tender, RWT Add on Convert, RSVR Merger Arb

## Introduction and Housekeeping - 00:00

I wanted to say happy Sunday, but it's certainly a hectic Sunday. Apologies for the sound issue — I got a new mic, and I always use two laptops on the call, and I think the call bounced from one laptop to the other, which resulted in no sound. You should have the email and all the files are up and running. If you can download the SSR Q&A PDF, that would be very helpful.

There are also several attachments to the email, and in the SSR Q&A tab the same. We have the **Morgan Stanley non-AI top picks Excel file**. We have the **Fidelity US fixed income** file. We have the **Standard Chartered "inching towards a Fed hike"**, which is now at 87% probability. We have the **Morgan Stanley 3Q update on equities and bonds**. We have **BlackRock, Why Japan Matters to the US bond market**. We have **JP Morgan on the Iran war, AI and transformation**. We have **Deutsche Bank on AI infrastructure and private capital**.

We have **two models on EVC and PRTH**, which are both small caps and are sensitive, with a higher beta to the market, but I think both very attractive long-term — one on a merger, one on a cyclical turnaround in advertising.

It's about 20 megabytes in attachments, which is probably why I'll have to resend the email to those who haven't received it — we'll do that after the call.

## Economic Calendar and Fed Rate Decision - 02:13

If you open to page one of the PDF, we'll start with the economic calendar. There's not much in terms of earnings next week, but we do have a pretty important economic calendar.

- **ADP**, the weekly employment change. While we had strong job numbers this month, they were mostly government, lodging and leisure, and healthcare related.
- **Empire manufacturing** is supposed to drop precipitously on the 15th, which is tomorrow, from **20.6 to 15**.
- **MBA mortgage applications**, then **retail sales on the 16th** — relatively weak; you are expected to see a bounce from **−0.6% to 0.8%**, which is still under 1%. **Ex auto and ex gas, that number drops to 0.4% because of the increase in gasoline prices.** Diesel prices are above **six dollars**, so that's a pretty big issue. As I'm talking, the Nasdaq is down about **1.1%**, so it continues to get a bit weaker.

Then after the retail sales data, in the afternoon, the **FOMC rate decision on the 16th**. The market is broadly **expecting a 25 basis point hike now because the war is looking to last at least through the end of this year**. And despite what Hassett and everyone says about interest rates, it's almost like the banks drawing their consensus higher are going to force the Fed — despite **Warsh and Waller being more dovish than the rest of the FOMC**. There are 12 voting members and it looks like they want to force the FOMC to hike. **If they do not hike, you actually could have the bond vigilantes forcing the 10-year even higher.**

## Market Risks and Fed Decision Impact - 04:19

This is a major issue for the markets. **If they hike, that's going to be very bad for commercial real estate, which is why we were adding a bank short that's the most exposed to commercial real estate in the United States. If they don't hike, the 10-year could go higher, which is going to be bad for tech. So you lose if you do, you lose if you don't.** This is what happens when you have stagflation — this is not exactly stagflation, but it's stagflation-like, where **the Fed has to hike into a slowdown because everything outside of AI in the United States is not growing as robustly** as it was over the last few years.

## Market Strategy and New Screener Tool - 05:07

But there's always something to do in this market. **Adding shorts and more eventy longs, like this Dutch tender; adding to Redwood, where I think the sell-off because of the convert was overdone; adding to some baby bonds there.** There are definitely things to do in this market — it's just not very obvious to most.

**Going forward, I'm going to do the ideas at the beginning of the call, to be mindful of everyone's time, and then focus more of the macro at the back end.** We're trying to redo this.

An update on the screener: **the screener will be out in October.** It has helped me with a number of new ideas that I have in the backlog — **that's how we have five live ideas on this call.** Going forward we're going to have a lot more ideas to discuss, **at different levels of conviction, in the mid-cap and small-cap space where you don't have everyone on Twitter talking about them.** The screener will also be available to **premium subscribers, silver and above**. It's on a different website, because it's linked to a SQL database and our current website doesn't have enough memory to run it, so it's going to run on our new website. Eventually the entire website will migrate over to something a little cleaner.

## Upcoming Earnings and Merger Arbitrage - 06:36

After the FOMC — where unfortunately **I was not expecting a hike, but now it looks like there's a high probability of a hike** — we have:

- **Philly Fed business outlook**, supposed to drop from **47 to 32** on the 17th.
- **Initial jobless claims**, which I expect to be flat.
- **Housing starts at 1.3 million**, still relatively low because of **mortgage rates at 7%**. They're supposed to bounce simply because they were down 12% in July, but after this I don't expect them to rise much more — **building permits are going to be down in August, from 1.43 million to 1.40 million**, so the following month you're probably going to see housing data fall back down again.
- On the 18th, **industrial production**, also very slow at **0.3% from 0.2%**, and the **leading index**, very anemic at **0.1%**, which tells you all you need to know about future growth expectations in this environment.

What I'm really worried about is **after the midterm elections**. With the tariffs against Canada, with the escalation of the war — if the administration doesn't care about the midterms and is simply willing to offer people **$5,000** of their own money back as some sort of a bonus, which would cost **$1.3 trillion** — the administration is quite dangerous, and **what could happen after November could be peanuts compared to what's happened going into the midterms.** Especially if they feel threatened, they might take some drastic measures, especially after the Supreme Court effectively said the tariffs are illegal. That could include the US **stopping export-import altogether, without tariffs, as threats** — with countries where we have deficits, we could temporarily just stop doing trade with 50% of our trading partners, which would be absolutely ridiculous, but I'm not ruling it out. Some members of the administration have already hinted at it — not that they're going to do it, but that it's a possibility.

## Upcoming Earnings and Merger Arbitrage (Continued) - 09:12

The good thing about the earnings calendar is there's really nothing happening, so we can focus on economics, on event-driven situations and on hedges. The only two important companies reporting next week are Trip Advisor / trip.com and **Lennar**, which is a home builder. **The only thing I'm going to be reading on the earnings front is Lennar, because I want to see how their sales are doing with higher mortgage rates.**

**Large-cap merger arbitrage.** The only one of these we own in size is **Warner Brothers–Paramount, which has compressed from over 20%**. **We own a little bit of NSC/UNP as a tracker, but not a lot.** Most of our merger arb exposure is in much smaller names.

**Roku / Fox.** There are antitrust issues around **Roku and Fox A, but that spread is just not wide enough to be involved in.** Several media reports last week had the **DOJ preparing to expand its antitrust probe into Fox's proposed $22 billion takeover of Roku**. Investors were slightly surprised but not shocked, given only 70/30 odds of clearing; they still think the base case is that it closes. Roku was well defended — multiple setters on the desk, a lot of big hedge funds putting the spread on after it widened. **It widened out to a 6.5% IRR to a February close**, then there was some buying at the end of the week because they think it will close. **That spread is not wide enough for us to make any money on, because I'm not going to borrow money to put on a silly spread like that.**

**Kinetik (KNTK).** Bloomberg reported Kinetik is going to sell itself, so **that's on the watch list. We haven't added it yet — we don't know enough about the situation.** For those who want to do work on it, the ticker is KNTK. It's a Blackstone-backed energy pipeline company — partially owned by Blackstone.

Kinetik Holdings Class A shares trade at **54.75**. On the holders list, **Blackstone owns about 14.8%, BlackRock 11%, Goldman Sachs 10%** — so these three big asset managers collectively own about **34%**; **Jamie Welch, the CEO, owns about 4%**. So almost 40% is owned by these three asset managers and the CEO, and the rest trades. It closed at 54.75 and obviously rallied last week. We don't know enough about this business to figure out where the takeout price will be, but **I have a call with Jefferies and one of the Goldman analysts next week to figure out what would be an appropriate takeout bid.** Based on my math, **the company trades at about 13 times forward EBITDA and about 25 times trailing EBITDA**, because they have a big margin improvement expected in 2026 — and that's what I'm trying to understand: are they going to get credit for it? **Their EBITDA more than doubled from 100 million to 250 million from the first quarter to the second quarter**, so I need to understand what drove that before I can determine the takeout potential. The stock's up **30%** over the past year because of that improvement in profitability — which followed a run-up in commodity prices after the wars in Ukraine and Iran, because this company owns **a large network of gas gathering and processing infrastructure in the Permian basin in New Mexico**. It benefits not just from throughput and storage — **it also takes a percentage of the price exposure**, which is more of a variable component. So we have to do more work on that and get back to you.

"Oxtay," we talked about. There's also the **Goldman Sachs TMT conference** from the week, which we'll touch on later. **ATAI — the deal was expected to close on September 11th, so that likely closed.**

**Air Canada odd-lot Dutch tender (deck p. 4).** On the following page is our first trade idea: the **Air Canada Dutch odd-lot tender**. We have a lot of Canadian subscribers. **If you're Canadian and not subject to Canadian withholding tax**, the ticker is AC on the Canadian listing. **It closed on Friday at 27.53, and this Dutch tender is supposed to take place on September 24th, between 29 and 34.** If you take 27.53 against 29, **the minimum you would make is 5%**; at the top it's **17%**. You'll likely get closer to the 5 than the 17 — but that's the range. **The repurchase capacity is about 28 million shares, about 10% of the total shares outstanding.** Based on the odd-lot provision — we've done these odd-lot tenders before — **there's priority allocation without proration for holders of less than 99 shares.** So if you own 99 shares across 10 accounts, each held separately at a separate broker, 99 × 10 × 27.53 means you can own about **$27,000** of this in that scenario — and if you made the most you could, 17%, you'd make about **4,600**. **So this is mostly for small traders, but it is something to take a look at.** The transaction is funded by liquidity proceeds from the recent sale of their Aeroplan loyalty platform, which they got paid cash for.

## Valuation Model Updates and Short Book Ideas - 16:37

### EVC & PRTH valuation model updates (deck pp. 5-6)

On the next page are the **EVC and PRTH valuation model updates**. We shared our thoughts on these situations over the past two weeks, and **one of our analysts has modeled them for you — the Excel is attached to the email and to the Discord.**

What we did: we calculated our own **cost of equity using the company's betas and the risk-free rate — we used CAPM, with the new 10-year Treasury as the risk-free rate and a risk premium of 4%** — and put in the company's cost of debt. **We then modeled out free cash flows from revenue to EBITDA to EBIT, adjusted for capex to get to free cash flow to the firm, forecast that out to get net present value and the terminal value, and subtracted net debt to get a fair price.** These are not perfect calculations, but **this implies roughly 59% upside on EVC.** You can read the overview we did two weeks ago on its fast-growing pivot into advertising away from its core business — it's been growing like a weed, and it trades relatively cheap to its cash flow growth.

**PRTH.** The CEO owns like 60% and is trying to buy the whole company at like **five times cash flow**. There are **two activists involved** — we talked about that about a week ago. **We bought this in the low fives, and we sold half of it above six.** It rallied because two small hedge funds got involved. **We can add back to it if it sells off on Monday back into the fives.** Long term, the activists say this is worth **$15 to $20 a share**, because it generates about **200 million of unlevered free cash flow** — after debt service it does about **100 million of free cash flow** — and if you value that 200 million, growing at single digits, it would probably be worth **about $2.6 billion, even penalizing the company with a 2% small-cap premium in its weighted average cost of capital. That gives us about $21 a share.**

That doesn't take into account AI risk, because it has a software component, or the other risks in this type of DCF beyond the small-cap penalty I added. The good thing is the company **provides payment and data processing services — it's linked to payments, which is less disruptible by AI** — but it does have a software component. **So it should probably trade at a discount to this $21, but it probably should not trade at 5.94, which is where it closed at the end of the week. It basically trades at a quarter of what its cash flow says it should.** The company could easily be a double if the CEO says: instead of buying it for six to seven dollars a share, **I'm going to pay nine to ten dollars a share just to get you guys off my back.** Or the board could say they want to sell the entire company.

Because the CEO owns so much, when he's buying out the rest of the public float **you basically need a majority of the minority**. Of the 40% of shareholders not aligned with the CEO, **the activists have 4%, so they need to convince basically another 16%-plus of the other voters to vote with them, and they can reject the CEO's offer** — then he either has to increase his offer or sell to someone else.

One reason these shares trade so cheap is **the CEO has been doing roll-ups of all these small companies — he made a $15 million acquisition last month.** People think: in the meantime he's making all these acquisitions, the cash flow is becoming less certain, and if we don't let him buy the company, is he just going to keep growing his empire with small acquisitions? That hurts the thesis a little. **He can make these acquisitions because he controls the shares — the majority-of-the-minority doesn't apply to acquisitions, only when he's bidding on the business.** So it's not a perfect trade, but the company trades so cheap that we did take a position. **We sold half at a decent profit — over 15% — and we will likely buy back if it sells off again.**

### Valley National Bancorp (VLY) — the CRE short (deck pp. 7-9)

**On the short book, we need more shorts because the market is quite volatile.** I expect the sell-off to probably get worse by tomorrow unless there's a magic deal announced in the morning. Already the Nasdaq's down about 1.25%, the S&P from −0.45% to −0.6%, the Russell −0.36%. **The 10-year has actually come down to 4.95% because of the fear in the market.** WTI is around **103**, Brent around **108**, both up about 3%; even natural gas is up 2% in sympathy. European natural gas is the highest in three years; US natural gas at 2.87 isn't very correlated, but it's still up.

So what are we going to do? **We think there's now a decent chance they are going to hike, despite Trump throwing a tantrum and threatening Warsh and the rest of the Fed.** We think **Japan is going to hike on September 18th**. **The ECB already hiked last week**, which is one of the reasons the 10-year ended higher, and the Bund and French 10-year were higher. **If the Fed hikes, the 10-year might actually fall because growth expectations fall — but commercial real estate is not valued off the back end of the curve like residential mortgage rates. Commercial real estate is benchmarked off SOFR, which is basically the same as the Fed funds rate.** So if Fed funds goes up, commercial real estate companies that have been stressed for years since COVID — office, shopping centers — are going to be exposed, and **banks that are overexposed to commercial real estate are going to be exposed.**

So there's a bank called **Valley National Bancorp, VLY**. **It's a short target primarily due to its structural exposure to commercial real estate debt.** It has **elevated liability costs**, a **very minimal NIM** — a smaller margin than its peers — and its valuation relative to risk: **it's close to its 52-week highs despite being one of the most exposed banks to commercial real estate**, and it didn't really sell off. Remember the 2023 banking malaise: this bank was **down 50% from January 2023 to almost $6 a share by April 2023**. Now it's basically trading at **five-year highs — it peaked around 15, the highest since 2009, and it's now back down close to 14, around 13.80.** **We think this bank could easily be down 20-30% on an interest rate scare.**

- **CRE concentration:** about **325%** of total risk-based capital in commercial real estate across office, multi-family and retail. That exposes the bank to refinancing stress — a lot of these loans will have to be refinanced and take markdowns, especially in urban office and retail. It isn't as risky as in 2023, when it had about **474%** exposure, **but the issue is it sold a lot of its good properties and kept a lot of the bad properties.**
- **High net loan-to-deposit ratio and funding costs**, which expose it to net interest margin erosion.
- **Very little fee revenue.** Unlike JPMorgan, Citi or Bank of America, which have a lot of fee income to offset interest-rate mark-to-market losses, **non-interest income — fees, wealth management, capital markets — is only 13% of Valley's total revenue.** That leaves the top line almost entirely dependent on loan spreads, with very little room for error.
- **Lagging credit-loss allowances:** bears highlight that Valley's **ACL has historically trailed regional benchmark peers** (the deck: ~1.19% of loans vs ~1.4-1.5%), leaving less cushion for unexpected credit losses — one of the reasons it's such an attractive short.
- **Short interest isn't too high — around 9%, about 42 million shares**, roughly seven days to cover. So the squeeze risk isn't super high, like the quantum names, and there isn't gamma-squeeze risk like Tesla. **This is a short I can put on and not be as scared about. If short interest were 15-20% I'd be more worried.** I think it could sell off **30-50% again** if we have another interest rate scare.
- **More exposed to wholesale funding markets** than super-regional peers, which are more susceptible to interest-rate spikes.
- **Total loan book about $51 billion; commercial real estate about $30 billion, almost 60%.** Non-owner-occupied CRE about **329%** of total risk-based capital. **It had a multi-quarter rally to near its 52-week high, so I feel more comfortable shorting it here.**

**Risks:** the 42-million-share short (~9% of float, seven days to cover) isn't small. Management has de-risked some of the CRE portfolio, taking exposure down from **476% of risk-based capital to 333%** — still one of the two or three highest in the country relative to its market cap. It expanded its **CET1 ratio to 11%**, basically to avoid rating-agency downgrades, which is why **S&P Global revised its outlook to positive — but I don't think S&P ever thought we would be hiking rates again.**

Book composition: **multi-family about $12-13 billion in New York, New Jersey and Florida**; retail commercial real estate **20-22%** — the most risky: grocery-anchored strip centers and neighborhood retail across suburban NY, NJ and Florida; **office 10-12% of CRE, $3-3.5 billion, about 6% of the book**; industrial/warehouse **10-12%**; healthcare/senior living **5-8%**; construction **5-7%**. The geographic exposure is on page nine.

### Exiting SK Telecom (SKM) on the AI slowdown (deck p. 10)

We'll jump to the next idea on page 10. **We're exiting SKM. We've done really well in this name. We're just worried about this AI slowdown and panic after Dario's letter being published.** We think he's going to **delay his IPO till October at the earliest — next year is the most likely.** And **OpenAI definitely cannot do an IPO this year.** So how are these guys going to raise money? Anthropic could do it privately; OpenAI less probably.

**If they're not raising more money, how are they going to support Microsoft, Amazon and Google's cloud backlogs? And if they aren't supporting the cloud backlogs, how are those companies going to increase capex? That's the flywheel I'm worried about. If they don't increase capex, who's going to buy Nvidia GPUs so that it grows earnings 30% every year, and who's going to be buying all this memory?** It's not going to matter in the next quarter or two — all this capex is still going to happen. But the market is hoping for 30% more AI capex next year and 30% more the year after. At some point it has to stop, and **a deceleration is enough for these stocks to sell off.**

That's what has happened: **the market multiple has compressed about 13%.** The reason the market is only 3% from all-time highs is **a 12% rally in the Mag 7** offsetting the decline in a lot of mid-caps — and because **S&P 500 earnings were up 30%, even ex the one-offs in Anthropic, OpenAI or SpaceX valuations**, and earnings are in the denominator of the P/E, that growth makes the S&P look cheaper. So even though the S&P is only down 3% from its highs, **it's actually down 13% in multiple compression.** But assume the Mag 7, which is what's rebounded, shows lower earnings growth — that's a negative trigger, because **half of their growth has been coming from these two companies, OpenAI and Anthropic.** I don't think their capex slows in the near term, but it could become a bigger issue later.

Over the weekend, **Anthropic CEO Dario Amodei publicly called for a formal, industry-wide slowdown of the pace of frontier AI model development.** The trigger was a security incident involving an autonomous agent swarm — the **OpenAI–Hugging Face incident** — which demonstrated dangerous emergent misalignment behaviors. **Anthropic unilaterally committed to letting external third-party auditors like METR vet internal training pipelines and model releases**, while urging democratic governments to enforce safety caps on capability leaps. **OpenAI's Sam Altman and xAI's Elon Musk publicly backed Amodei's call.** A deliberate pause on frontier scale-up caps short-term revenue acceleration assumptions across the sector, hitting high-valuation private multiples. Instead of endlessly chasing raw parameter scaling, there could be a capex slowdown. The market is now forced to transition from capital deployment into immediate enterprise monetization — and for a company targeting a **$1-2 trillion valuation ahead of a draft S-1**, any voluntary brake on capability deployment introduces immediate multiple-compression risk for private secondary markets.

**Why I'm exiting SKM specifically:** this was a **high-asymmetry deep-value proxy trade — a dividend-paying Korean telecom holding 3.7 million shares of Anthropic, giving massive private exposure relative to its $15 billion market cap. The arbitrage is now closed.** **At its lows of 19 to 20, where we started to look at it, SKM was fundamentally dirt cheap, providing Anthropic optionality for near free. Now it's rallied to the 38-40 range.** The stock was already pricing in a **neutral core telecom at $11 billion plus Anthropic at a private mark of $1.7 trillion.** With SKM near its 52-week high, the market had fully credited the bulk of the upcoming IPO upside. **So it makes sense to sell SKM completely — and if you need to buy it cheaper, you can always buy it cheaper. I doubt Anthropic is going to be worth more than $1.7 trillion anytime soon with all this.** I'm making a note in the alerts that **we're selling in the pre-market near highs, as Anthropic is fully valued in the valuation.** The shares are at 38 on the close; even if they open a little lower, I'm not worried about it.

### Did someone yell "slow"? (deck p. 11)

Someone shared this thought process with me, and it provides a more balanced approach to the slowdown news over the weekend.

**Sam Altman and Dario Amodei both chose this week to talk publicly about slowing the AI race, but the market read-through is very different depending on which one you're listening to.** Altman said that given everything happening with safety, **right now would be an ill-advised moment to go public** — we all know he couldn't go public anyway, because his growth had slowed. Dario, meanwhile, argued that better pacing would allow labs to do the necessary safety work **without sacrificing commercial advantage**, while adding that progress would **still seem fast**: "we must pace the frontier."

**Altman can afford to sound cautious because OpenAI is not trying to price an IPO next month — but Anthropic is. So you have to take Dario more seriously.** That is what makes Altman's comment awkward for Anthropic-linked sentiment. OpenAI was not expected to list this year anyway — its own CFO had already pointed employees toward 2027 — so Altman loses very little by declaring this an ill-advised moment. **Anthropic's investors, meanwhile, are pushing hard toward an October 26th IPO at a reported close to $2 trillion valuation.** You can see the problem: apparently this is a dangerous enough moment that going public is ill-advised — unless, of course, you happen to be the company trying to sell $2 trillion of equity into it. That doesn't change Anthropic's revenue, compute demand or fundamentals, but it's hardly the framing you'd choose heading into a roadshow. **Negative for IPO sentiment, potentially negative for pricing**, and another reminder that AI-safety rhetoric gets a lot less theoretical when someone is asking public investors to write a very large check. The stock reaction on Monday will tell us about positioning and how over their skis investors are, particularly with the macro so poor.

**Dario's letter is different. It's not a growth warning — he goes out of his way to tell you that.** There is nothing in the letter about delaying model launches, cutting compute, reducing capex or lowering revenue expectations. He does not touch the numbers at all. So fundamentally there's very little here. **Sentiment-wise, though, the fact that he felt the need to explicitly reassure investors that pacing does not mean commercial slowdown tells you he knows exactly how the market is going to hear it.** When the CEO has to explain that slowing down does not actually mean slowing down, investors will probably spend at least a few days debating what "slow" means.

For the actual AI trade — **Nvidia, memory, power, optical and the broader infrastructure complex — nothing here changes the near-term capex number. No hyperscaler guide has moved.** Nobody has delayed a data center or pushed out an accelerator order outside of the power issues you already know about. **For now, this is sentiment risk, not fundamental risk.** Still, two of the people running the labs the entire AI trade is benchmarked against have publicly introduced some version of a slowdown just as $1-2 trillion valuations are being underwritten on the assumption that capability growth keeps compounding. That's enough to create short-term chop and multiple pressure even if the numbers stay untouched. **It becomes a fundamentally different story only when pacing turns into an actually delayed model, a pushed compute deployment, lower accelerator demand, or a company guiding lower citing AI spending. Until then, do not confuse uncomfortable headlines with a capex cut nobody has actually made.**

**In my opinion, the real risk is that Anthropic or OpenAI run into funding issues and slow down cloud spending, which lowers the backlog growth of Microsoft, Google and Amazon, who in turn cut capex. But we won't know that for a while. In the meantime, the market will correct a bit to price in the uncertainty.** So that's a balanced approach — his view and my view.

### Redwood Trust (RWT) — adding on the convert overreaction (deck p. 12)

**We added to Redwood equity and baby bonds on a stock overreaction to the convert issue last week.** They issued a **$150 million convert — not to raise new money, but to refi a 2027 bond due next year.** I think this was poorly advised given how weak the mortgage sector was, but they offered the convert **at a 35% premium, which is close to book value** — so they thought they weren't diluting people below book — and **a 7% coupon, much lower than their other bonds at over 9%**, so they thought they'd save interest.

**Convertible arbitrage funds, though — because this was only a ~$400-odd million market cap — had to short equity to hedge the $150 million convert. Assuming an initial delta between 50% and 65%, the arbitragers had to short between $75 million and $100 million of common stock**, which would have been almost impossible to do in a day. **With a ~$500 million market cap, dumping $75 million of short selling into the open market in a single session creates immediate downward pressure. So the stock was down like 22%.** They also took a **2% book value hit — very low — by selling $190 million of their worst assets in a legacy bridge loan sale.** Two percent of book for selling that much is not bad, but it spooked the market. **Redwood then said it would buy $20 million of its stock concurrently, at a big discount to book.**

**We started to add the stock when it was down about 18%, at 3.41 and 3.45, and then it continued to fall to about 3.20. We added another 10 bps at 3.21, and brought our average below 3.60. Then we added five bps to the baby bonds** (the deck: RWTQ and RWTS) **at yields of about 11%. These baby bonds will likely be repaid at par — just like it just repaid its '27 bonds at par.**

**UBS** published a note on the 10th, before the sell-off, when the stock was at **4.26**; it traded down almost to **3.21** — about 25% from the report — but they published after the convert was issued. They said the transaction addresses a portion of the 2027 maturity, reducing near-term refinancing risk and improving financial flexibility; the company intends to repurchase **$20 million** of stock, which should offset some of the convert holders' delta hedging; Redwood still has about **$111 million** of capacity remaining under its $150 million repurchase authorization as of June 30th, 2026; net dilution remains subject to final conversion pricing. **"We believe today's sharp equity sell-off reflects a combination of convertible arbitrage hedging, dilution concerns, and uncertainty around the ultimate amount of 2027 debt that will be retired, rather than a materially deteriorating credit profile… we believe the magnitude of the stock decline appears disproportionate to the underlying economic impact."**

Obviously the stock also sold off because mortgage rates were going up, so **you can't have too big a position here, because interest rates continue to be a risk. But we've been adding, and if rates go higher and it sells off further, we will continue to add, because this company survived COVID.** It was one of the best-performing REITs during COVID; it was the least levered, and it's still one of the least levered. **I've personally talked to the CEO several times, and I think he understands the risk. His customers on the mortgage underwriting side have very high FICO scores, around 750 — so it's not a default risk, but there's a mark-to-market risk because of interest rates.** **It's not like UWMC, where you have trash management that did an equity rights offering and diluted the living crap out of investors after taking billions of dollars in dividends for themselves.** This is a completely different type of management team. But again, maybe they surprise.

### Reservoir Media (RSVR) — merger arb under a $10.50 bid (deck p. 13)

**We added RSVR after a PE offer at $10.50 a share, after the shares weakened below $9.** We added in the low nines — we weren't able to time it perfectly. **This stock was trading over 11 because of the offer made at the beginning of the year; people thought there'd be a topping bid above 10.50. The shares then fell all the way to nine. We bought in the low nines, and we think that at minimum this $10.50 should go through.**

**The $10.50 cash bid from a majority insider, Wesbild / Richmond Hill, and competing interest from an activist fund that also wants to buy, Irenic Capital, keeps this deal in play.** Wesbild controls about **44%** of the equity and Irenic owns **9.2%** — combined, **53%** of the stock. It moved quickly because of that: it has a small float, and a lot of the stock is owned by private equity firms and hedge funds. **There's no financing condition on the takeover.**

This is a highly profitable company: about a **$624 million market cap, a billion enterprise value, $450 million of debt, but $70 million of EBITDA** — so leverage is moderate. It does **$40 million of free cash flow after paying interest, about an 8% free cash flow yield.** It's generated **$30-50 million of cash flow consistently over the last four years**, profitability has been growing, it's not a very cyclical business, and it's grown **10-18% a year over the last five years**. It's a **music media company — it distributes music recordings, serves customers worldwide, and has top-line revenue of about $180 million with a 65% gross margin, because a lot of the business is royalties.** We think these bids are real. **Before the company received them, the unaffected price was around 7.50.** The company has grown since, because the offer was in February. **Worst case, I think it trades down below eight; best case, it trades above 10.50 because there's a bidding war. But we think there's a high probability this bid goes through.** Its catalogs generate **high-margin, inflation-hedge recurring revenue from streaming and performance royalties with low correlation to macro cycles** — it's not a traditional advertising company — so for that reason it should trade at a higher multiple.

### Morgan Stanley top mostly non-AI stock picks (deck p. 14)

On page 14 you have **the Morgan Stanley top mostly non-AI stock picks — all mid- to large-cap names. This is for you to screen and look for your own ideas.** The ideas are across LatAm, Europe, Asia and the US, weighted outside North America because of how high valuations are here. To give you a flavor: **ABN AMRO Bank** in the Netherlands — the Dutch economy is still doing okay — overweight at $48 with modest upside of 10%; **ADES**, a Saudi Arabian energy company, with 40% upside; **Abu Dhabi Commercial Bank**, down because of the war, with about 20% upside; **Accton Technology** in Taiwan, about 86% upside — a developer of networking products, adapters, hubs, switches, routers — down about 25% from local highs, one of the few names on here that is data-center related. They have **Alibaba** with 60% upside, **Alpha Bank** in Greece, and among the higher-octane names **Contemporary Amperex Technology** in China.

**We think there's going to be a major stimulus announced in China.** There's a big China equity research report in the SSR Q&A tab — **Franklin Templeton, Mapping China's AI Ecosystem**, and **Morgan Stanley on China's next industrial revolution**, which wouldn't fit in the email: a ~200-page report on China robotics, capex, equity fundraising, capital markets, industrial innovation leadership and automation. They haven't been as stimulative as in the past, but given the global slowdown, **if they announce a stimulus this could be a very interesting year for Chinese equities.** The Chinese 10-year is basically a 3% spread to the US 10-year, and I think the market's expecting some stimulus already.

You also have some dividend aristocrats. **Coca-Cola — MS thinks 13% upside despite the GLP-1 fears.** It has a decent dividend yield that should be defensive. **KO closed at 82.29, with a dividend yield of about 2.4%, but it's a growing dividend, and free cash flow generation of about $12-14 billion a year on a ~$370-379 billion market cap — take $13 billion on $379 billion and that's a 3.5% free cash flow yield. So it's not glaringly cheap**, but it has sold off pretty dramatically from local highs in the low to mid 90s — not as much as I would have liked, but more than the S&P. Over the next week I'm going to go through all these names and see if one or two stand out. **But there are some value traps, like Lloyds Banking.**

There are some **defense-correlated names like Rolls-Royce, which I think are interesting** and we've talked about in the past; some more **eventy names like Warner Music Group, WMG, that could eventually be sold**; **Northrop Grumman** on the defense side; and **Palo Alto, which we also like, is on the list.**

### Iran war update — the East-West pipeline goes offline (deck pp. 15-19)

**Brent oil surged over $100 a barrel — over 107 in futures on Sunday — after Saudi Arabia halted some energy operations near its border with Yemen due to attacks**, highlighting supply risk from the region. Saudi production was already declining.

- **Saudi Arabia temporarily shut down the 1,200 km East-West pipeline following targeted drone strikes originating from Iraq and Yemen.** Satellite imagery identified damage and black smoke near the pipeline's path south of Medina. **The pipeline had been Saudi Arabia's primary overland bridge to bypass the blocked Strait of Hormuz**, moving crude from the eastern fields to the western port of **Yanbu** on the Red Sea.
- **The GCC meeting in Oman on Monday has been postponed** — Bahrain was the first to say it wasn't going. **That's one of the main reasons oil is up 3%; it was only up half a percent before.**
- **Volume offline:** the pipeline shutdown, though temporary, **removes up to 4-5 million barrels a day of capacity.** Global oil markets are dealing with an estimated **5.7 million barrels a day** of total supply offline across the Gulf from pipeline closures and the maritime blockade. Before the shutdown, western-coast loadings at Yanbu had risen to **2.9-3.7 million barrels a day** as flows rerouted away from Hormuz.
- **Bab el-Mandeb:** ships are still crossing — **about 26-28 vessels per day, down from pre-crisis averages of about 70.** The **Houthis seized strategic positions along the Yemeni coast and Perim (Mayyun) Island directly inside the strait**, declared safe passage for general commercial traffic while actively targeting **Saudi-affiliated vessels**, forcing mainstream non-shadow tankers to turn off AIS or bypass the Red Sea via the Cape of Good Hope — **which is why tanker rates are so high, $800,000 a day, the highest in history, because of the ton-miles around Africa.**
- **Hormuz:** the largest structural disruption to seaborne oil since the 1970s. We were supposed to have a peace treaty, but six months later there's no peace. Since the conflict between the US, Israel and Iran broke out in late February 2026, the choke point has remained largely closed. **IMF PortWatch data: transits have collapsed from a pre-crisis average of 85 vessels a day to 5-10.** They were higher last week but are down again. The few tankers still moving are almost entirely Iranian-flag (fewer now because of the US blockade), Chinese shadow-fleet tankers and dark-fleet hulls with AIS off. Pre-crisis, **20-21 million barrels a day** passed through Hormuz, about **20% of global consumption**. With Iraq, Kuwait, Qatar and UAE exports severely throttled, **over 10 million barrels a day remains shut in or trapped**, and the East-West strikes take **an additional ~4 million** of bypass capacity offline.

**Because Chinese imports had fallen, that kept oil under 100 — but now I think if you see further escalation in the Red Sea, oil could easily go to 120.** On **September 13th, Iran signaled intent to present Gulf nations with a bilateral agreement reached alongside Oman to establish a restricted temporary shipping lane** — but with Iran retaining unilateral veto over which vessels cross, and now that meeting has been delayed as well. **The market was already indicated lower — Nasdaq futures down 95 bps, and now we're down 1.2%.**

Midweek clashes pushed oil to **99** before the Yanbu attacks. On Wednesday (deck p. 17) **Iranian forces fired missiles at US warships, including an aircraft carrier, at least three times in the past week, after the US bombed three Iranian VLCCs.** The IRGC claimed strikes on eight tankers and two vessels, with **UKMTO confirming one cargo ship hit off Al Faw in Iraq**; the US did not retaliate for Wednesday's attacks. **White House aides are now bracing for a years-long conflict — Vance, Rubio and others have told Trump Tehran could hold out under the blockade, potentially extending the conflict past January 2029.** Trump himself said Wednesday the war will end immediately after the November midterms — very convenient for him to say — and promised voters **$5,000** if Republicans win, which would cost roughly **$1.3 trillion**. **US gasoline averaged about $4.22 a gallon Wednesday, from $3.19 a year ago (AAA).** On Thursday, **Iranian strikes hit a US military installation in Jordan**, severely damaging one **A-10** (a wing detached) with about **eight F-15s** suffering minor damage; **US air defense fired roughly 30 Patriot interceptors, which are incredibly expensive — which is why we have to spend half a trillion dollars replenishing interceptors and missiles.**

**Saudi production was already down before these attacks: 6.2 million barrels a day in August reported to OPEC, down about 1.9 million from July and the lowest since 1990**, due to US-Iran hostilities, Houthi embargoes and strikes near Yemen. **Exports fell by a third to around 3 million barrels daily**, pushing Brent above 105 and WTI over 100 with 4%+ gains in one session. OPEC averaged 38 million barrels a day overall and cut its 2026 demand growth forecast (the deck: to 380,000 b/d from 580,000).

On the Iran-GCC meeting: there were originally signs of progress, which is why **oil fell 2% on Friday**, but now we're back up. Before the meeting was cancelled, **Oman and Iran were driving an initiative to secure GCC buy-in for a temporary agreement: inbound Persian Gulf vessels routed through Iranian territorial waters and outbound traffic primarily through Omani waters.** Tehran and Muscat hoped a unified GCC endorsement would pressure Washington into lifting its blockade on Iranian ports; the GCC wanted a pragmatic mechanism to restore passage. **The issue is US veto risk: diplomats cited by the FT say Hormuz cannot fully reopen without US buy-in, and Washington remains opposed to Iran asserting unilateral transit control or collecting navigation fees.** **We think oil could rise to $100-120 a barrel in short order if both straits are fully closed.**

**The other side — Robin Brooks** (deck pp. 18-19). Oil out of the Persian Gulf is averaging around **15 million barrels a day, so we're running at 75% capacity** versus before the war. Based on standard price elasticities of demand, that implies **a 33% premium over pre-war prices, so Brent should trade around $95** — though he wrote this before the East-West pipeline blew up. His argument: the crowd that pushed for $200 oil in March and April has gone quiet, and the passage of time makes such a spike increasingly unlikely — **market psychology** (the apocalyptic forecasts were so wrong), **the narrative shift on Iran** (the blockade exacting a heavy toll), **learning-by-doing on supply chains**, and lower Chinese imports — so **Brent around $90-100 adequately prices the disruption, and we won't see $125 again.** **I believe oil should be higher than 100 right now — closer to 120 — but for a lot of these controversial topics I like to give you two views.**

## S&P 500 Internals and the Magnificent Seven - 01:03:04

**Why is the S&P 500 only 3% from all-time highs?** Consider everything thrown at the market in this holiday-shortened week alone: **Brent +7% to over $100; diesel above $6 a gallon for the first time ever** — which will lead to inflation, because trucks use diesel to ship our goods; **the two-year up nearly 30 bps to 4.6%; the 10-year up nearly 20 bps to 4.96%; the odds of a September hike from 60% to 89%; and 2026 hike pricing from 1.4 to two** — not to mention the AI apocalypse doomerism. And for all those wrenches, **the S&P 500 was down less than 1% on the week. For the S&P, macro pain really has been an illusion.** The resilience is even more impressive against a tough seasonal backdrop — September weak in a midterm year — and complacent positioning and sentiment: low volatility, low demand for downside protection, overweight positioning, stretched futures, bullish sentiment readings.

**Despite the index being down just 2.2% from its all-time high, there's been a large deterioration beneath the surface.** After Thursday's close, with the index down 3% from highs, **a mere 17% of names were trading above their 20-day moving average and only 33% above their 50-day** — down from 70% a month ago. **These metrics are far closer to an internal washout than you'd normally expect with such a mild index drop.** **The S&P can thank its 34% weight in the Mag 7** — the average stock, as measured by the equal-weight index, is down a lot more than 3%, but **the Mag 7 has rebounded from its June swoon: the 12% rally since the start of August has made up for broad-based softness.**

The other source of resilience is earnings. **Both 2026 and 2027 EPS estimates are up 16% year-to-date, pricing in 32% growth this year and 14% next year** (ex the Anthropic, SpaceX and OpenAI valuation gains). That growth has allowed **the S&P to tolerate a 13% decline in its forward P/E and still deliver a 12% year-to-date return.** The 13% de-rating comes from multiple sources:

- **Higher rates** — the low in rates in 2025 coincided with the peak in valuations: **above 23× then, closer to 21× now.** A further lift in yields will put more pressure on valuations.
- **Deteriorating earnings quality, mostly for the Mag 7** — its forward P/E has dropped **30% since last October**, to the lowest since 2025's Liberation Day lows and close to the 2022 tech bear-market lows, capturing now-poor cash generation and higher capital intensity. **I still think there's downside to the Mag 7 if Anthropic and OpenAI growth drops precipitously**, but we'll see.
- **Peak growth rates** — some segments over-earning or pulling forward earnings power. **2Q26 was likely the peak quarterly growth rate at over 50%, and 2026 the peak annual rate at 32%, with a sharp deceleration in 2027.** The market sniffs that out and ascribes a lower multiple.

## FOMC Rate Hike Probability and Market Resilience - 01:06:44

How much longer can we count on huge positive revisions to dull the macro pain? **With 2027 EPS at $416, implying 14% growth on top of stellar growth this year, we're not sure how much more realistic upside there is. Take 7,600 on the S&P divided by 416 — that's 18× forward for 2027, which doesn't look that expensive. But these growth estimates might be cut after this AI slowdown.** That's why the market is so uncertain — stuck in this area, because 18× looks cheap for next year, but if growth slows, is 18 the real multiple? Is it 19? 20? Of course, taking the under on earnings has been precisely wrong all year, which is why analysts keep raising forecasts until clear evidence challenges them — so estimates could keep rising into year-end even if challenged by reality in 2027: one-time gains not repeating, tough comparisons, less upside to hyperscaler capex. We still see potential for more volatility through the rest of the month and into October; rate dynamics can't be ignored, since a further climb and tighter financial conditions could weigh on earnings growth as well as valuations.

**The FOMC hike probability is now 87%** (deck p. 22). The move in CME FedWatch has been driven by three factors. **First, CPI was hotter than expected, with energy bottlenecks — August CPI came in at 3.4% y/y, 0.4% m/m.** Core CPI did meet expectations, but the headline was hot, driven by energy and supply disruptions tied to the Middle East conflict and Red Sea bottlenecks. **Second, a resilient labor market — August payrolls added a stronger-than-expected 162,000 jobs with unemployment steady at 4.1%**, giving the Fed leeway to focus on inflation.

## Japanese Yen Intervention and US Treasury Market - 01:11:00

**Third, after Chair Warsh's hawkish tone at Jackson Hole and subsequent signals from Fed governors, desks including UBS, Citi and Macquarie rapidly revised terminal-rate forecasts from extended hold to two hikes before year-end** — the Fed's first increases in three years, taking the target to **3.75-4%**, with the market pricing higher-for-longer through 2027, which is not the best for valuations, gold, silver or crypto. **However, I personally think this is going to be one and done. After this hike, the back end will be range-bound, and with slower growth globally and hopefully an ending war, we'll get back to cutting sometime in 2027. But if the war extends, that gets thrown out the window.**

**The yen and the Japanese bond market** (deck p. 23): Bessent tried to support them and then failed. Speaking Tuesday at Southern Methodist University, Bessent highlighted his coordination with Japanese officials on recent **joint yen purchases — the first US involvement in many decades** — and his pressure for BoJ hikes. The yen strengthened to about 153; right now it's at 154. **The Bank of Japan has spent about $85 billion of reserves selling Treasuries to support the yen**, and I believe they're blowing through reserves to keep it here, especially with oil rising. It peaked around **163** on July 28th, they intervened to 157, it went above 160 again on September 1st, then Bessent and the BoJ brought it down to 155, and it's been in the 153 area for the last few days. Yen shorts have been hovering near extremes. Bessent basically said "I'm the house now" — when we intervene I have insight into what the BoJ will do; bet against me if you want.

**The Treasury buyback** (deck pp. 24, 36-37): when he accelerated long-end buybacks, **$6 billion a month isn't going to do anything to a $40 trillion Treasury market, or ~$10 trillion of long-dated Treasuries.** **The 10-year is at 4.96% over the weekend, a three-year high near the psychologically critical 5% threshold — a classic bond-vigilante rejection of Bessent's intervention strategy.** Dealers had built up expectations for a **"bazooka" buyback of over $10 billion a month**; when Treasury capped the expanded 20- to 30-year purchase at **$6 billion and only executed $5.19 billion due to a lack of sellers — banks didn't want to sell paper at those prices — the market marked down the implied value of the Bessent put.** Buybacks recycle dealer balance sheet into new auctions, but a $5-6 billion operation is a rounding error in a $30 trillion-plus market and doesn't reduce aggregate duration when net issuance is near record highs.

**Then we had a failed 30-year auction: $22 billion tailing significantly at 5.308%** — institutional buyers refusing to absorb long-end duration without a steeper premium. Then the hot headline CPI and energy spikes. **The 30-year made new highs even after a good auction — 5.34% last week** — and the two-year also spiked.

**The Russell 2000 is the most sensitive to rates, with 40-45% of the index unprofitable** (deck pp. 25-26) — which is why the Russell was so weak last week. Pre-2010, non-profitability was only 15-20%; the surge to over 40% reflects years of zero rates, early-stage biotech inclusions and software growth names. **By comparison, less than 7% of the S&P 500 is unprofitable, and the S&P SmallCap 600 has an explicit positive-earnings screen keeping its unprofitable share around 20%. So if you want small-cap exposure, you should buy the S&P SmallCap 600 index over the IWM, and short the IWM.** The Russell is more rate-sensitive for three reasons: **30-45% of Russell 2000 debt is floating-rate, benchmarked to SOFR — i.e. Fed funds — versus less than 10% for the S&P 500**; a **2026-2027 maturity wall**; and **40% of Russell 2000 constituents have interest coverage below 1.5×** — their EBITDA cannot cover interest payments, which makes sense because they're not GAAP profitable.

**PPI met expectations, but the headline was higher on energy** (deck p. 27): **+0.4% m/m in line; 5.4% y/y vs 5.3% expected; core +0.2% m/m vs 0.3%; core 4.6% y/y in line.** Core came in cooler, showing pipeline pressure outside commodities isn't spiking out of control yet — which gave the Fed some room before CPI. **CPI came in hotter on the headline, but it looked like the market had priced a slightly hotter number, as the S&P rallied 80 bps on 9/11** — there was also talk about the GCC meeting with oil down 2% — and by the end of the day we started getting bad news, and over the weekend we've given it all back.

**BofA on CPI:** beyond its implications for next week's Fed, not particularly informative about the underlying trend. The upside surprise was driven largely by core services, +0.3% m/m; within that, **wireless telephone services surged 5.9%, contributing roughly 10 bps to core — likely AT&T retiring certain unlimited plans and raising admin fees, a one-off** — and airfares and lodging away from home added a combined 8 bps; airfares in particular may have been boosted by higher oil. These should prove temporary. **What's slightly concerning is core services ex-housing rose 0.5%, the highest since January and up 3% on the year.** Core goods was very low at 0.11%; shelter low at 0.26%. **The irony is that the weight of CPI items running above 3% fell sharply — which is what Warsh talked about at Jackson Hole.** By category, **fuel oil +52% y/y, gasoline +27%, airline fares +23%**, and the next one was tobacco at only 6% — so it's driven mostly by fuel. Month over month: fuel oil +10%, gasoline +4%, airfares +2.7%, and the next biggest only 1.1% (motor vehicle maintenance and repair).

**Hammack is hawkish** (deck pp. 32-33) and I'm sure a couple of other Fed officials are, but **she's not as important as Waller and Warsh. The problem is that with this war lasting over a year, they might not have a choice but to agree with Hammack.** Waller holds a structurally more powerful position: he's one of the seven Board governors, appointed by the President and confirmed by the Senate for 14-year terms with a permanent vote, whereas Hammack is CEO of the Cleveland Fed, a rotating voter. **So whatever Waller and Warsh want is going to carry more weight.**

## Hedge Fund Deleveraging - 01:33:45

Kashkari's position is on the next page. **A lot of betting markets are betting on a September hike** (the deck: Kalshi at 80%, up from 53%).

**Diesel prices are a major issue going forward** (deck p. 34). Diesel goes into consumer goods prices, and **diesel is now at all-time highs, above the 2022 Russia-Ukraine surge — about $6.05 a gallon.**

**Michigan consumer sentiment** (deck p. 35) dropped from **51.7 to 47.8 against an estimate of 51**, driven by the expectations component; the Iran war and oil prices are likely playing a role, with **one-year inflation expectations jumping from 4% to 4.6%.**

**Energy was the only sector higher at the end of last week and the only one with positive breadth** — the eighth time this year a single sector was green, and six of those were energy alone. Higher energy prices are pulling equities back toward the immediate post-Iran playbook: **semiconductors and energy regaining leadership as the broadening trade fades. The thing I'm worried about is that semiconductors could sell off and energy might be the only sector that rallies tomorrow because of this AI slowdown** between what Dario said and what Altman said. The grind higher in oil is firming inflation expectations and driving yields up, weighing most on cyclical and rate-sensitive stocks. **Brean Capital calculates the daily correlation between oil and Treasury yields at about 0.85 — the highest monthly average since the Iran war began, if it holds for September.** Fundstrat notes **the equal-weight S&P has broken its uptrend** as discretionary, healthcare and financials weaken, while semis and the Mag 7 prop up the indexes. Nomura flagged a **semiconductors-and-energy barbell** as one of the highest-Sharpe baskets — but that could break tomorrow. Call skew has been rising in both. **The Situational Awareness guy was buying calls right before this capex slowdown — he's probably going to lose a ton of money on Monday, since he was buying naked calls.**

**Treasury buybacks are not working.** Bessent learned an important lesson on September 9th: **once you show markets you're willing to adjust policy when they move against you, they will always demand more.** Treasury had already doubled longer-maturity buybacks, then Bessent **tripled "Operation Twist" to $6 billion — and bond yields went up** on the day he announced. With the buyback fervor fading, traders will refocus on the more important cross-asset threat: **a broadening commodity inflation impulse.** The 10-year moved toward 4.85% that Wednesday as traders had positioned for something larger; a solid 10-year auction steadied things — **the market is absorbing supply, but it increasingly requires higher yields to do so.** None of this addresses the structural pressure: heavy government borrowing colliding with a corporate calendar swollen by AI capex, all competing for global savings. **The average G7 10-year moved above 4% for the first time since 2008 (BMO).** Buybacks are a band-aid for a wound that needs surgery.

And the inflation backdrop has worsened: Brent above 100 alongside historically tight product markets won't show in this week's PPI and CPI, but there'll be an eventual pull higher on core as commodity prices work through. **Copper at an all-time high broadens the move beyond energy; agriculture is joining, with higher diesel and fertilizer costs from Hormuz, tighter Black Sea grain flows and the risk of a powerful El Niño.** There's a timing wrinkle — headline reaches core with a lag, and coming BEA revisions may mechanically suppress near-term core PCE, so the next prints may look calmer than the backdrop. **Energy, metals and agricultural price increases are creating a 2027 core inflation problem.**

**BofA on T-bills:** $7 trillion of bills outstanding, most maturing within a year. **Assuming the Fed hikes 75 bps this year — BofA is too aggressive in my opinion** — annual interest cost on T-bills alone would rise by **$50 billion, about 15 bps of GDP.**

**Midterms** (deck p. 38): Trump promised at the Republican midterm convention in Dallas **$5,000 for every adult US citizen if Republicans win both the House and Senate. We have $40 trillion of debt; this would add $1.3 trillion in a single day. I'm still waiting for my DOGE dividend — where is it?** The DOGE dividend was $5,000 funded by efficiency cuts that never happened; the tariff dividend of $2,000 was pushed away by the Supreme Court; ACA "savings" of $1,000-plus checks from expiring subsidies; and now $5,000. **Across all of them we've been promised over $13,000 of rebates — basically $2.5 trillion — none of which I personally think we'll see.** The rest of the world is watching and concerned: **despite China's growing trade, its share of US Treasuries outstanding has fallen under 2% for the first time in 25 years.**

**BoJ hike coming September 18th** (deck p. 39). **Some think it could be 50 bps; I think it'll be 25**, to **1.25%**. Governor Ueda will aim for a balanced tone at the September 17-18 meeting, two days after the Fed, to limit yen and bond volatility in either direction. A dovish read could weaken the yen again — benefiting exporters and PM Takaichi, who favours easy policy, but adding inflation pressure; the Bank wants to avoid a return toward 160. A hawkish read would reinforce the yen's move to 153 and lead markets to price a higher terminal rate. The Bank wants long rates to reflect market views but worries a surge could undermine the recovery.

**The German far right is gaining power.** Their elections are a couple of years off, but it's something to watch — the coalition may revisit retirement at 63, which I think would lead to protests.

**Hedge fund deleveraging** (deck p. 40): **we've seen significant hedge fund deleveraging this summer. There was some re-leveraging after Situational Awareness, but after two or three weeks we found out Citadel sold 80% of what it bought, and now we're seeing deleveraging again.**

## Nuclear Sector Initiations and Outlook - 01:34:01

**Jefferies initiated on nuclear last week** (deck p. 41) — I'm surprised they didn't have a view before. **Buys on BWXT, CCO (Cameco's Toronto line), DML, EU, MIR, NXE, SOLS and KAP; Holds — basically sells — on GLO, LEU and UEC.** Laurence Alexander sees **~$55 trillion of nuclear-related capex through 2100** — I don't know how he can forecast that — hinging on AI, drones, climate, security and replacement: data-center demand, broader electrification for physical AI and robotics, and climate stress on aging nuclear plants and refineries, for a long cycle with frequent fly-ups. **He favors the integrated producers — NXE, CCJ and KAP** — and global capex via MIR.

- **Cameco** is the West's largest listed integrated uranium company — mining, conversion and reactors via Westinghouse; the thesis hinges on global nuclear capacity growth and the upcoming contracting cycle; operating discipline and a strong balance sheet make it a prime beneficiary of the West's build-out given full vertical integration.
- **Kazatomprom (KAP)**, state-owned, is the world's largest uranium producer with some of the lowest costs; a structural supply deficit should drive realized prices higher and EPS growth over the decade, with first-quartile costs, supply discipline and one of the highest realized-price sensitivities to spot.
- **Mirion (MIR)** is uniquely positioned across nuclear instrumentation, radiation detection, dosimetry and reactor monitoring — low direct uranium exposure, defensive growth tied to fleet utilization and new builds; upside drivers are SMRs and large-scale new builds, fleet life extensions, advanced-reactor monitoring and defense programs.

The ratings across integrated miners, developers and services/equipment are on page 41.

## Goldman Sachs TMT Conference Highlights - 01:36:47

The Goldman Sachs TMT conference (Communacopia + Technology) is the biggest technology conference every year for those not familiar. **AI was the clear thread, but the debate has moved from model capability to infrastructure scale, power, ROI and real enterprise adoption.** Nvidia and SpaceX framed AI compute as a new industrial asset class, while Shopify, Capgemini and Goldman's CIO pointed to agent-led workflows as the next software paradigm. **Cybersecurity stood out as one of the most direct AI beneficiaries** — Palo Alto and CrowdStrike argued for more automation, model diversity and platform consolidation, and **Palo Alto's CEO said "the AI security industry has not even been built yet"** (the two are ~$490 billion of market cap versus AI companies in the trillions, and cyber was the number-one theme at the conference). Semis remain tight and constructive; T-Mobile pushed back on satellite disruption.

**Bottom line: AI demand is broadening across the stack, but investors still need proof that capex converts into durable revenue, margins and cash flow.** This was before the Dario comments. TMT datapoints were broadly positive, yet investors remain hesitant to fully recommit to the AI infrastructure trade — bottom-up indicators constructive (tight supply, robust token demand), top-down concerns around volatility, rates, geopolitics, policy, the law of large numbers and Chinese models. **The central debate is whether AI capex converts into durable profitable growth: hyperscalers need to turn backlogs into sustained enterprise AI revenue, which is now under question; customers are shifting from "token maxing" to ROI-led "token optimizing"**; constraints across power, memory and data-center capacity should keep the market tight until 2028.

- **Nvidia:** AI is a fundamental shift from retrieval computing to generative computing; **the bottleneck has moved from chips to power**; demand is broadening rather than narrowing.
- **SpaceX:** increasingly sees itself as an integrated AI infrastructure platform, not just a launcher/satellite operator (the deck: CFO Bret Johnsen, another hosting deal worth ~$1.11B/month from December 1, on track for $100B ARR).
- **Adyen and Shopify:** Adyen's weakness appears driven less by PayPal's comments than by US hedge-fund concerns that **Stripe Link** and agentic shopping agents could disintermediate it; **that bear case feels overstated** — agentic commerce will likely run on multiple rails and, as Shopify noted, simply become another channel. **To me, the PayPal issue seems more PayPal-specific**, as Mastercard, Visa and Shopify said trends remained similar to Q2 while PayPal cited tariffs on its China-Europe business. Shopify positions itself at the center of AI-driven commerce with Sidekick and Autopilot.
- **Palo Alto and CrowdStrike:** AI is already transforming vulnerability detection — tools like **Mythos** find and remediate issues much faster — and **a multi-model strategy is essential**, since different models uncover different blind spots; AI still struggles with edge cases and context, **which is why you still need these cybersecurity companies.** Palo Alto sees platform consolidation; CrowdStrike highlighted its new **SafeLine** platform against autonomous agentic attacks.
- **Capgemini and the GS CIO:** AI as a shift from IT services to business transformation — you can read that.
- **Lam Research (LRCX):** CFO **Doug Bettinger struck a very bullish tone, calling the industry "fundamentally sold out"**, with leadership more concerned about missing the upside than managing a downturn; 2027 WFE growth DRAM-led, leading-edge foundry/logic closing the gap, NAND solid, **cleanroom availability the key constraint through the end of 2027.** STM and T-Mobile you can read.

## Blackstone Real Estate and Broader Market News - 01:41:28

- **The head of Blackstone real estate quit after a year**, given the real estate interest-rate issues and the private credit issues.
- **Palmer Square — one of the largest CLO platforms, with $27 billion of AUM — is looking to sell its business, likely at the top of the credit cycle.**
- **Tanker rates hit an all-time high at $800k per day. We still own IMPP. We think the stock could be 50-100% up from here, even if tanker rates come down from these local peaks.** Tanker rates were up **43%** just in the last week. **Refining capacity is the lowest in modern history, which is why all the refiners have been rallying.**
- Dan Niles' post-Labor Day commentary is in the deck (p. 48).
- **The FAO food price index rose sharply in August, led by sugar, but wheat, cereals and vegetable oil are rising too. This is a red flag.**
- **The Trump administration is seeking to rescind a rule put in place to stop private equity funds from bribing public officials, which is absolutely absurd** (p. 50).
- Market recaps (pp. 51-54): **Palantir and Nebius** sovereign AI partnership; the **EU and Canada** reportedly set to announce a major trade partnership after Trump's tariffs; **BlackRock's bitcoin ETF** with $3.7 billion of inflows this quarter; **OpenAI's CFO saying the consumer business is accelerating — and then all of a sudden they wanted to slow down their models**; **Schwab retail selling software and buying semiconductors**; **LME copper at a record $14,533/ton, up 47% in 12 months** on data centers, renewables and grids, with stockpiles concentrating in the US; **Meta launched Muse**, its personal AI agent; **IREN's 2 GW Sweetwater hub conditionally included as base load in ERCOT Batch Zero**.
- **The US government is backing Rigetti and D-Wave**, resulting in a mini short squeeze.
- **Uber's president and COO purchased 70,000 shares for $5.3 million** (spoken as "one of the biggest insider sales"). **We've talked about Uber being cheap.**
- **Apple announced iPhone Duo, its first foldable iPhone** — which is why Apple started to rally — but it's quite expensive at over $2,000.
- **Nebius** said AI infrastructure demand is still running well ahead of supply; **Google** plans **$13 billion** in Finland AI infrastructure over two years, its largest-ever European investment; **Amazon** is tapping the UK bond market for the first time with sterling bonds; **Anthropic researcher Jacob Coxon**, a 27-year-old former OpenAI employee, is leaving the AI industry over concerns the race is moving too fast.
- **The Bank of Japan now owns 46% of all JGBs. The US is going to start increasing its bond ownership much faster, just like Japan.**
- **White House advisors privately warned Trump Iran may resist US pressure through January 2029.**
- **US mortgage demand fell as rates moved higher — the average 30-year back around 6.85%.** Trump said the Iran war will end immediately after the election, which makes no sense — it should end before — so you can't really trust that at all.
- **Nvidia and Palantir expanded their AI partnership.** PPI came in slightly hotter than expected.
- **Microsoft plans to triple its global data center capacity from 12 GW to 38 GW by 2032.**
- **Oracle delivered more than 300,000 GPUs to AI cloud customers since the end of Q4, tripling Q4's capacity delivered; RPO surged by $209 billion to $664 billion — but at the same time Larry Ellison announced a new 10b5-1 plan to sell $7.5 billion of stock.** Oracle also guided FY revenue to at least $90 billion and adjusted EPS to ~$8, while disclosing **$28.5 billion of capex, negative $5 billion of free cash flow and a $20 billion ATM equity program — which means it's going to be issuing stock. It's probably why Larry Ellison is trying to front-run that and sell his own stock — first time basically ever. That's not going to be good when the market opens.**
- **Adobe** reported Q3 revenue slightly better than expected.
- **Uber's Dara Khosrowshahi also bought 141,000 shares for $10 million at an average price of $71 — his first open-market purchase since May 2022, when he bought 200,000 shares at $26.73, and the shares have almost tripled since. I think the market was quite excited about the two big insider purchases of Uber.**
- **SpaceX's CFO said the company signed another AI compute hosting deal that will generate $1 billion of revenue per month.**
- **GameStop's CEO bought 1 million shares at an average price of $20.38. Don't know what he's doing.**
- **Robinhood** ended August with **28.6 million funded customers, +7% y/y**, and total platform assets near $400 billion, +8%.
- **Saudi oil output fell to its lowest since 1990.**

## Q&A Section - 01:50:22

We're going to pause here and go into Q&A — quite a long call.

## Impact of Fed Fund Cycle on Agency REITs - 01:50:30

*"How does the Fed funds cycle and long-duration yields impact agency REITs?"* (Bitcoin Tina)

**When interest rates go up, agency REITs' cost of capital goes up and their NIM goes down. Eventually the bonds they buy will also see higher rates, but in the near term their cost of capital goes up first, because they have a lot of short-term funding. So it shrinks their margins and they should fall. Agency REITs should fall as rates go up.** But my view is that **rates are going to remain in this 5% area. I think we'll have one hike, that hike is going to slow down growth and cause fear, and unless the war extends and gets a lot worse, rates will be in this 5% area. But if they're not, then you shouldn't own agency REITs.**

## Reservoir Media Deal Closing - 01:51:20

**Because a lot of these REITs pay double-digit dividends, even if they fall 5-10%, the dividends will offset some of that decline. So I'm not panicking about them yet.**

*"When is the Reservoir Media deal expected to close?"*

**There's no close date, because it's not clear which bidder is going to win yet.**

## Interesting Baby Bonds with 10-Year at 5% - 01:51:41

*"In addition to Redwood, what other baby bonds look interesting with the 10-year at 5%?"* (Blake)

**I'll work on that for this week.**

## Larry Ellison's Oracle Stock Sale Cancellation - 01:51:49

*"Just FYI, I believe Ellison cancelled his sale."* (Brad)

Thank you, Brad. **Oracle chairman Larry Ellison just cancelled the planned $7.5 billion stock sale without selling a single share — especially, I think, given this AI slowdown. I don't think he wanted to look like he was panicking.**

## Analysis of McDonald's Stock - 01:52:10

*"Thoughts on McDonald's?"* (Yojo)

**McDonald's, just like Coca-Cola on the Morgan Stanley list, is a defensive stock. It sold off because of GLP-1 concerns, and it sold off with interest rates. It was at 340 on February 27th at the annual peak; it's now down to 252 — about 26% — and it pays a dividend of about 3%, and it has been growing its dividend forever, just like Coca-Cola. So I would say Coca-Cola and McDonald's are both defensive names to add. McDonald's has obviously been more affected by the GLP-1 crisis, whereas Coca-Cola is still doing well because Diet Coke sales are still doing well.** McDonald's is trading at the lowest level since July 2024, and October 2023 before that — **the bottom end of its three-year range.**

Deutsche Bank published on fast food last week, talking about GLP-1 taking some same-store sales lower. McDonald's brought back **Spicy Chicken McNuggets** with its Mighty Hot Sauce in the first week of September, is launching a **SpongeBob / One Piece Happy Meal** nationwide on September 15th with 15 collectible toys, is bringing back the pumpkin and crème pie, and has an in-app deal for a **$2 breakfast sandwich** weekly. **So it looks like they're desperate to bring sales back** — and you're seeing the same across Starbucks, Freddy's, Caribou Coffee, Blaze Pizza, Cold Stone, P.F. Chang's, Red Robin, Wetzel's, Pret A Manger, Little Caesars, Potbelly, 7-Eleven — all offering lower-priced items to attract customers.

**Wells Fargo** published in August after meeting McDonald's CFO: modest upside of 9-10% because of ongoing headwinds; digital offers back, a marketing pivot — **90% of customers who lowered Q2 purchase frequency have been given app offers.** McDonald's is still the leading US fast-food operator with sustainable margins and mid-single-digit global growth, but growth is slowing, so they use **15× EBITDA versus a historical 20× target**. **Their downside case is $260, base case $300, upside $390 — published a month ago. So McDonald's is already trading about 4% below their downside case, with about 19% upside to the base case.**

**I'm not an expert in McDonald's. It is a defensive name like Coca-Cola. I would be wary of some of the GLP-1 issues — I think a sixth of Americans are on GLP-1s. But it's a defensive name overall. I'm not that excited personally. I'm not going to be buying McDonald's or Coca-Cola, but I do think they are more defensive names.**

We're going to close the call; the recording will be up shortly. **S&P futures are down 45 bps, the Nasdaq down 1.22%; the Russell is only down 16 bps, because with the growth scare, bond yields are actually back down to around 4.96% — not up despite oil being up 3%.** We'll see if that changes tomorrow. I hope you have a great trading week.

*Call closed at ~1:53.*