# Special Situations Report — Weekly Call (readable notes)

- **Title:** PCE, FOMC (Warsh's debut), Iran peace breakdown, adding gold via Alamos Gold (AGI) special situation, real-estate dividend adds (DX / NLY / RWT), FPH land special situation, ServiceNow cheap, SATS/SpaceX unlock hedge, Salesforce buybacks, data-center delays
- **Show:** Weekly SSR research call (premium subscriber recording — no public video)
- **Guest/host:** Jay Singh (founder, Special Situations Report; ex-Goldman Sachs)
- **Date:** 2026-JUN-21 (Sunday call)
- **Source PDFs (this folder):** `transcript.pdf` (premium transcript), `report.pdf` (written SSR), plus supplemental research (AGI gold model, ServiceNow model, Joel Greenblatt special-situations class, Standard Chartered / HSBC / DB / BlackRock notes)
- **Note:** readable notes auto-extracted from the premium PDF; section headers carry the recording's `HH:MM:SS` cue. No public video, so the per-name table has no deep-links. Wording otherwise verbatim from the recording.

---

2026-06-21 - SSN - PCE, FOMC, Iran Peace, Adding Gold AGI Special Sit, Real
Estate Dividends DX, NLY, RWT, ServiceNow Cheap, Arb Discussion

Introduction and Market Futures - 00:00:00

Okay, happy Sunday, everyone.

Just like, uh, the last 110 days, there's been a back and forth around the Iran deal. Uh, it looks like Iran backed out of
Switzerland, um, after new threats and, uh, continued attacks to Lebanon. Um, but that will be very small part of our
discussion on today's call, but I did want to mention it given that futures are down 40 bips for the S&P 500, 37 bips for the
Nasdaq, and 55 bips for the small-cap 2000, the Russell.

If you look at commodities, oil prices are up 2% on WTI, up to 78, uh, from 76 last week. Uh, Brent is now 81 spot 40, uh,
from below 80 last week. And it looks like there's just, you know, there's no peace, um, in the near term. It doesn't look
like, um, the two sides are working together and that $324 billion bribe of $24 billion in in, uh, frozen funds and a $300
billion construction fund, which no one knows, uh, who will fund, weren't enough, uh, to get Iran on board.

So, um, welcome back to, uh, to chaos here. Uh, it looks like Nasdaq is now down more than half a percent. We'll see how
things open up tomorrow morning, uh, and if anyone backs down by then.

Overview of Today's Call and Company Earnings - 00:01:35

So, um, what we're going to cover on today's call, um, we're going to cover, um, company earnings and the economic
calendar. As you guys know, Micron is reporting next week, the most important company. We also have FedEx, Carnival
Cruise, NovaGold, Paychex, uh, trip.com.

As you all know, um, we are long TripAdvisor. And despite the fact that they sold a business, um, last week, it's about half
the company's market cap, as BK I'll mention. Um, the stock barely moved. You know, it was up maybe 35 cents, uh, to
about $13 a share. Now, you know, our cost base on the shares is something like $10. Um, so we're still up 30%. Um, from,
you know, it was, it was the lows were around nine.

Um, but, you know, the planned sale of TheFork, um, you know, for about 300, for for about 700 million, uh, in an all-cash
transaction is, um, is quite a good sale. Um, so that was announced on the 15th. Um, and I believe American Express um,
agreed to, to buy the, to buy the division. Um, and what might have happened is, you know, that takeover may have
already been rumored or somehow priced into the stock.

But what we'll do is, you know, in the merger, uh, arbitrage channel, I will put in a summary, um, of TheFork transaction. I
need to just confirm when it's going to close. Given such a small deal, I don't think there's going to be much antitrust risk.

So really Jeff Smith at Starboard Value has been pushing for a sale. Um, so I'll also just copy this and put this in the, uh,
SSR Q&A tab so that we can continue to monitor TheFork sale. You know, I do think that TripAdvisor is like 50%
undervalued, but, you know, while it's up 30% from our cost, you know, I think it's, it's probably worth over $20 a share.
We'll see how the other divestitures play out.

Gold and Real Estate Dividend Stocks - 00:04:14

Um, we're also starting to become more positive on gold, um, given, you know, a very sharp sell-off, um, from the local
highs down to around 4,100, uh, dollars an ounce. We think that central bank buying is is frankly not going away. And, uh,
the dollar's been strong because of, uh, the flight to safety.

But, um, you know, gold, as you guys know, back in January got to around 4,000, 5,420. And now we're at 4,000, so 4,150,
so 4150 divided by 5420. Yeah, we're down about 25% in gold in a period of a few months. So, we think it's interesting and
one way that we would like to play gold after the earthquakes, uh, last week for Alamos Gold, um, is to buy this, you know,
mid-cap, um, Alamos Gold play. Um, its main asset, which is about 60% of NAV, had nothing to do with the earthquakes.
Furthermore, no one died, no one was injured. Um, so we think the impact of the stock should have been about 4-5%, it
sold off 25%.
Um, the stock, the US shares, we bought the Canadian shares because of Juneteenth on the on the 19th. Um, we are
buying the, um, US shares as well when the market opens. We have a 10 basis point tracking position. Um, we'll buy the US
shares which will likely open around 30. We think upside based on where gold is today is around 50, uh, but if gold rallies,
obviously the upside could be higher.

Um, we're also going to be exploring some real estate dividend value names. You know, we're last week as you know, we
added, uh, Dynex, uh, Redwood, NLY. You know, NLY is a 13.5% dividend, Redwood's around 14% dividend, Dynex is
around a 16% dividend. So high dividend names that, you know, given we think that rates have are near peaking if this Iran
deal can can close.

FPH Small-Cap Real Estate - 00:06:23

And then the FPH small-cap real estate play. So this is a California-based um, residential real estate company that has a lot
of JVs. You know, some of its JVs have dividended over 100 million to the company over over the past few years. It's trading
at a very depressed value, around 40% of book value. We think over the next three years it could be a 3x, um, if they're
able to develop some of the the high quality property that they own.

Um, it's a smaller cap name, so less liquidity.

Software Sector (ServiceNow, Salesforce) - 00:06:53

And we'll talk about ServiceNow. Um, we think that ServiceNow has, you know, about 80-90% upside. It's probably one of
the better software names and less AI exposed. Um, it's not as good as the cybersecurity names, which we all know, which
we owned and and they've all doubled, including DataDog and CrowdStrike. Um, but we think that ServiceNow is in kind
of the second category of of software names. And it continues to be depressed, um, just given the IGV ETF has seen
outflows, but overall, we think it's better positioned than the average software company.

We'll also talk about, you know, Salesforce, which is more of a software benchmark. They've announced 55, they've done
$55 billion of share buybacks, but the market has just ignored it. It shows you how software is just out of favor. We're not
advocating, you know, a buy of this company, but it just shows you despite tens of billions of dollars in share repurchases,
the stock has still been been in the doldrums as, you know, hedge funds have been selling software to buy
semiconductors, probably one of the most crowded trades of all time.

Okay? It doesn't mean that it still may not work for for, uh, the next few weeks and maybe a few months, um, but it is one
of the most crowded trades of all time.

Sats and SpaceX IPO - 00:08:07

We will also then talk about, um, Sats and and the SpaceX IPO. Um, you know, we think that at-the-money put buying,
which you can actually track, uh, you know, thousands of contracts every day, um, because you can't actually buy puts if
you're an institution that has, uh, an allocation to the SpaceX IPO, that's keeping that arb around over 40%, given the, uh,
upcoming August unlock for SpaceX. You know, so what we've done is we've, uh, hedged our Sats exposure by shorting
SpaceX, and it actually worked really well on Friday because SpaceX sold off more than Sats.

Um, and we've also have some covered calls on Sats. So what we did was we shorted 30% of SPX versus Sats and then we
sold some very juicy covered calls through October for like $10 a contract on Sats as well. Um, so most of the sell-off in the
SpaceX complex, you know, was absorbed by the SPCX short and the the call premium, uh, leaking that we sold.

Upcoming Economic Events (FOMC, BoJ, ECB, PCE) - 00:09:12

Um, we'll also discuss the FOMC. You know, we have a new Federal Reserve Chairman, Warsh. Um, we'll discuss the Iran
peace timeline. It went from a peace deal to now, um, Iran and, uh, B.D. Vance leaving Switzerland.

We will also discuss the Bank of Japan, uh, which hiked rates to the highest since 1995 and the recent ECB hike. We'll
discuss PCE expectations for the USA next week, which is the most important data point, uh, on the economic calendar
next week. We'll also discuss data center delays, which have been confirmed by Jefferies and JPMorgan, but semi-analysis
has challenged the thesis. They're not saying that there aren't delays, they're just saying the delays are not 70% like
Jefferies and JPMorgan are saying, but maybe, you know, something under 50%, which is still quite bad.

And then we'll discuss OpenAI and Altman getting a body blow from Microsoft, which, uh, has basically said that they're
going to use usage-based pricing and diversify their AI models, uh, outside of OpenAI and Claude. You know, you've seen
a number of banks balk at the price of Claude. Um, we've seen Uber have to cut back their AI spend because they blew it
all in the first quarter. Um, Microsoft has also been sued by a Michigan pension fund alleging that they lied about their
cloud profitability. So, you know, there are a lot of early signs that AI is getting overheated. We'll talk about those.

Um, in terms of the economic calendar, the most important economic events next week are the US manufacturing and
service PMIs, new home sales, PCE, personal spending, GDP, durable goods orders, and UMich sentiment. Also, as you
know, Micron is reporting this week, will be one of the most important earnings prints this week given how the
semiconductor memory segment is just completely, um, you know, at a blow-off top.

Special Situations and M&A Deals - 00:11:04

In terms of special situations, after jumping, um, you know, 14% uh, on Thursday, shares of, uh, Lionsgate were down 6%
on Friday following a report um, that Netflix is not interested in the media company.

Um, with respect to Roku, uh, Netflix also indicated that it did not make a bid to challenge the existing Fox bid. As you
know, Fox bid, um, for Roku, which really hammered, uh, Fox stock, uh, but they're trying to integrate, you know, that
advertising flywheel business.

The 2% spread in in the Webster Bank-Santander deal is is likely to, you know, likely to move tighter after the deal
received, uh, OCC clearance on Thursday afternoon. Uh, the approval will be seen as a positive signal that the review
process remained independent and objective. So that spread is gone. The deal still requires the European Central Bank
and the Fed, uh, to sign off.

In other news, the ESPR cleared HSR, OGN filed the definitive proxy. There's an article about Caesars potentially selling its,
um, AC assets, um, in New Jersey, and several ZIM-related headlines regarding a bid from Heim Skall.

Also, Huntsman and Olet announced an all-stock merger of equals on June 16th. The M&A terms imply an at-the-market
valuation and a 54%, 46% hunt equity split. The deal brings, uh, 400 million in operational synergies on a 1 billion EBITDA
base. Um, you know, a lot of chemicals names have sold off this year, so this is a way for them to to extract synergies and
preserve some valuation.

The widest spreads in the market right now are, uh, NSC, EUNP, and Warner Brothers-P-sky. Um, despite receiving
antitrust clearance, the Warner Brothers deal still has a very wide spread, which we think is quite attractive. Um, the
Warner Brother twists and turns continued on the 15th of June following a Wall Street Journal report indicating that the
Department of Justice staff was leaning towards recommending a lawsuit challenging the deal on the grounds that the
combination of the two movie studios would be anti-competitive. This is different from, um, the news earlier, um, in the
week that said that, um, you know, the DOJ had closed its investigation. Um, originally, um, the market had viewed that as
favorable, but now there, you know, there are rumors that, you know, they're going to sue. That's why the spread is as
wide as it is.

Um, you know, but I'm assuming after some divestitures, they'll get clearance. Um, in other news, CCRN filed the definitive
proxy for July 16th vote. EXTA received clearance from CADE. LLPX received German FCO approval, and Valaris filed their
transaction with A-Triple-C.

You can see Lionsgate, you know, rallying on the 16th and then selling off after Netflix said they wouldn't buy it. It's been
bid up for the last several months given that they have a big content portfolio that someone is going to buy eventually.
Um, Roku, uh, shares were halted on the 15th after the company announced they had received a definitive agreement
with Fox Group to be acquired for $160 a share for um, for cash and stock, valuing the company at 22 billion, which I was
quite surprised by. Fox will pay about $96 a share in cash and .9693 of Fox A-shares for every Roku class A and B share.

In other news, Honeywell Aerospace, ticker HONA, will begin trading. It was a spin-off from Honeywell on a when-issued
basis, um, on Friday, um, for a two-week period. After that two-week period, it will trade, um, as an independent stock. Uh,
again, that ticker, uh, is HONA. So within two weeks, that'll trade as a separate company.
Two Harbors, um, announced on the 15th that UWMC did not submit a proposal during the waiver period that Two
obtained to engage directly with UWMC in a potential transaction. As you guys know, they've been challenging this deal
and it hasn't closed, which is why Two rallied after the initial UWMC sell-off, given UWMC has sold off so much because of
its own problems. Um, so details of all these deals, the Webster deal, um, Lionsgate, um, bid fake, OGN, uh, the
Huntsman, Olin deal, um, the the Warner Brothers, um, DOJ clearance and then, you know, rumors of the lawsuit, CCRN,
Honeywell, um, HONA spin-off. All those details are on page four.

Bearish Positioning in Gold and Alamos Gold (AGI) - 00:16:20

On page five, um, we'll quickly discuss bare positioning in gold. Um, we think that bearish positioning in gold reached, um,
its highest level since 2017, according to Goldman. Um, as a result, um, we added some exposure, maybe a little bit too
early ahead of the Fed, but we added more exposure after the Fed meeting. Um, gold is down 25% from the January highs.
We added PHYS and GLD for spot ETF exposure. We also added miners, GDX and SIL, and a new, uh, miner special
situation at Alamos Gold, starting with the Canadian shares and then USD shares on Monday.

Um, we bought about 10 basis points to start, so about .1%. We'll likely double that on Monday morning. Um, shares were
down 19% in Toronto, the most intraday since 2020, after the miner trimmed its second-quarter production guidance as a
result of earthquakes that damaged its infrastructure at its Young-Davidson mine, which is one of its smaller properties.

So, this is a cheap, mid-cap gold miner that went through a temporary setback due to an earthquake outside of its main
asset. Um, its main asset is called Island Gold, and it's over, um, 60% of the company's NAV. Um, Alamos Gold, ticker AGI,
has a market cap, um, of about 15 billion dollars, so it's not a small company or a speculative company by any means. Um,
it doesn't have a lot of debt either. Um, so it's a relatively unlevered business, um, you know, with a total of, you know, as
net cash of 500 million. Um, so we think that on a go-forward basis, it's a relatively good risk-reward.

Um, if gold stays flat or rises, we think this name could rise about 65% to $50 a share within one to two years. Um, and we
think that the US shares, if they match the Canadian share sell-off, will open around $30 a share on Monday with upside to
50 over one to two years. Um, the company also has a very strong growth profile as well, so it deserves a higher multiple.
Right now, it trades at a 25, 20% discount to peers. Um, more than a 20% discount to peers.

If you look at its peer group, um, you can see, you know, it trades at a discount to Lundin Gold, Agnico Eagle, um, Royal
Gold, um, um, on a on a forward FY1 and FY2 basis. On FY2, um, it trades at about 4.7 times forward EBITDA, um, which is
cheaper than most of the companies in the index, which which trade at a median of about, uh, 6.1 times. Um, on a P/E
basis, um, on a FY2, it trades at eight times earnings, um, given its growth profile.

So on Friday, Stifel cut its price target on the company in Canadian dollars, uh, from from 70 to 75 from 80, which still
implies a lot, you know, almost a double. Um, you know, why did it downgrade? Because the two earthquakes, um, and
storm-related damage to a power line for one of its smaller properties resulted in three, uh, days of unplanned downtime.
So the company now sees 2Q production between 130,000 and 135,000 ounces, which is at the lower end of guidance.
Um, mining rates at Young-Davidson are expected to average, uh, around 5,000 tons a day, which is a couple thousand
below expectations. Um, Island Gold, which is about 60% of NAV was not affected and production for the second quarter is
expected to be in line with plan. It's underground mining rates, um, have increased to a record above 1,500 tons per day.
It's on track to increase to 2,000 tons per day by the end of 2026.

The 19% intraday drop, uh, looked like to be a severe overreaction by the market driven by automated trading algorithms
and a near-term panic rather than a fundamental degradation of the company's long-term net asset value. The
operational reality shows the impact of seismic events is localized and the stock sell-off, we think, is fundamentally
decoupled from the actual underlying damage to the business.

Um, you know, the Q2 production is only dropping about 12%, uh, at the midpoint down to, you know, 130 to 135,000
ounces from the baseline of about 151,000 ounces, so not terrible. Um, you know, like I said, there are zero injuries, zero
deaths. Um, there is an infrastructure damage that physically blocks access to two specific high-grade, uh, stopes within
the, um, the Young Mine complex.
In terms of the full-year drag, because underground mining rates at Young-Davidson are capped at 5,000 tons per day for
the rest of 2026 to implement extra ground support, full-year consolidated production will slide to below the low end of
guidance, um, at around 570,000 ounces for the year versus, you know, the original range of 570 to 650. The 19% drop
wiped out roughly 3 billion in market cap in a single trading session, um, versus actual damage to the company of under
100 million.

In terms of the cash flow math, you know, the company will miss out on roughly 20,000 ounces in the second quarter, um,
representing a deferred or lost revenue of about 46 to 48 million. Um, even factoring elevated all-in sustaining costs for
the remainder of the year, the total financial impact should be under 100 million. Um, structural cushion. Um, Island Gold
is the company's main asset, which represents over 60% of the NAV. As noted in the Stifel and Jefferies updates, Island
Gold is completely unaffected by the seismic events. Um, underground mining rates also hit a record, and this asset
should continue to grow. Um, the newly acquired Magino mill throughput is already approaching over 10,000 tons per day.

Um, so how much should the stock be down? You know, we think it should be down about 4 to 6%. So it was probably
overdone by about 15%. So the event and the gold sell-off, I think make this quite an interesting hold. Um, you can see
Jefferies shares that view with a base case up 40%, upside up 80%, downside down 15%. Um, their note is on page nine
and page 10, which lays out their gold production forecast, the company's all-in sustaining costs, you know, the the
multiples they believe the company should trade at, total revenue by year, um, total gold sales, uh, real costs, operating
cash flow, uh, free cash flow, and, you know, assets, liability, and valuation multiples at the bottom of page 10. And that,
um, you know, reflects, um, you know, this note was was written around the time of the, uh, of the guidance reset, so
these are all updated.

FPH Small-Cap Real Estate - 00:23:25

Um, another interesting name that we wanted to discuss is the FPH. Um, so we added, you know, to high dividend names
like Dynex at a 15.6% dividend, Redwood at 14 and NLY 13 and a half last week. We'll continue to add to real estate
names. You know, it looks like the market is going to be soft at the open. We'll continue to add to these names. Um, it
looks like, um, you know, despite, um, the rally in gold, you know, US 10-year is is, uh, is around 4.51 percent. So, up
slightly, but nothing too dramatic, except maybe four bips.

Um, you know, a name that provides more torque but it's a lot riskier, um, that we think could be a potential triple is a
company called FPH. Uh, it's a company called Five Point Holdings. Um, so obviously our RWT, NLY, and Dynex positions
will be much bigger, but this is a a small allocation that we think could do well over time.

Um, there's a hedge fund called Black Bear Value Partners that originally, um, shed light on this name. We have no
affiliation, but the market basically treats this company as this as a standard home builder, but in reality, it's a huge land
owner. Um, and as it develops this land or sells the land, uh, through JVs, there's a lot of cash that could be unlocked. Um,
it's a master plan computer uh community land developer trading at a massive discount to its liquidation value, and it
shouldn't trade like a home builder.

Um, based on the 1Q 2026 financial disclosures and the structural tailwinds of California real estate, um, you can basically
identify, um, the three big flagship communities that drive a lot of the company's value. One, uh, is called Great Park
neighborhoods in Irvine, California, which is located near the heart of Orange County, which is where most people have
been moving, uh, from greater LA. It's where the affluent people live in California. This is FPH's crown jewel. It is a
premium multi-decade development surrounding a 1,300 acre public park. Home sites here command massive pricing
power from public home builders like Lennar, Toll Brothers, etc. on the premium side.

Um, then they have some properties in Valencia, which is also in Los Angeles County. This is a massive, fully permitted,
master-planned city designed for approximately 21,500 homes and millions of square feet of commercial space. Valencia
is a cash-generative driver for the company through its JVs.

San Francisco's Shipyard and Candlestick Point. Uh, this is a long-duration massive urban waterfront project slated for
roughly 12,000 homes. While subject to long-term development delays, its terminal, uh, land value is immense due to its
geographic scarcity.
So, the reason why this company trades at such a big discount is that the market doesn't know how to value the joint
ventures. As highlighted by Black Bear Value Partners, FPH's most meaningful cash-generative asset is held in an
unconsolidated joint venture, which is called, which is the Great Park venture, which is, you know, the the land that they
own in Irvine, uh, in near Orange County.

And because this asset is owned in a JV, um, the market doesn't give them, you know, the right, um, or the fair fair
treatment for that ownership. And, so if you look at the company's financials, they look kind of sleepy, right? While FPH
can report quiet quarters with minimal consolidated revenue due to land sale timing, um, such as we saw in the first
quarter of 2026 where they had a $5 million loss on $13.6 million in revenue, which, you know, looks, you know, looks bad.
Um, they routinely extract massive lump-sum distributions when they finish developing properties within those three
projects, uh, including a record $231 million in distributions and incentive comp, um, in a single year recently.

So, when you calculate, you know, what FPH is truly worth, if you look at the trailing P/E ratio, it doesn't look that cheap
because the land sales are intentionally lumpy. So, the only way to value this company is by calculating a NAV. So, the
market cap is about, you know, 750 million. On the, so it's a very small name, and it doesn't have a lot of volume. It only
trades it trades a couple hundred thousand shares per day. Um, I'm sorry, the market cap, I'm sorry, the enterprise value
is 750 million, not the market cap. Um, but its accounting inventory is about 2.5 billion. So its estimated real estate NAV is
between 2.2 billion and 2.5 billion. And so, if you look at the, um, you know, the company has two share classes. So when
you look at the market cap at 370 million, that's not the true market cap. They also has B shares, so it has 72.4 million A
shares, and then it has 76 million B shares. So in order to calculate that that true market cap of about 750 million, you
need to include, um, both classes of shares. And, um, I will just, you know, paste that screenshot so you can see their two,
um, share classes in the SSR Q&A tab.

But when you account for those, you know, you basically get that there's 750 million of of, um, of equity value and the total
real estate's worth about 2.5 billion. So it's roughly a 3x. Um, FPH's accounting book value sits at roughly $12 a share. So
the stock trades at about a a .4 book value. Means you're buying, you know, the most premium Southern California land at
about a 60% discount to book. But then when you mark their thousands of unsold acres, uh, to their current market value,
um, you should get something like 15 to 17 and a half dollars a share. Um, so it's an asymmetric trade that will take place,
it will take a number of a couple years for this to trade to its fair value in my opinion, but, you know, where it trades at $5
versus its ultimate, um, fair value around $15 is quite interesting.

ServiceNow Deep Dive - 00:30:04

Um, then ServiceNow. So ServiceNow is on page 13. You know, for ServiceNow, you know, the question we ask is, how
often do you see a company grow free cash flow by about 36% in 12 months while its valuation falls by 60%? That's exactly
what's happening with ServiceNow. It has a healthy 7% 2027 estimated free cash flow yield, which is much higher than
that of the S&P 500 now. It, you know, we added to shares at about 95.48 after hours, post-FOMC on on June 17th.

Um, so for those of you who don't know what ServiceNow does, it is a workflow software platform that is a cloud-based
solution that helps enterprises and organizations across public and private sectors digitize workflows. The Now platform
helps business leaders realize value from these investments by incorporating advanced technology into the flow of work,
end-to-end, across the enterprise for every department and persona. The company has approximately 8,400 enterprise
customers, so very diversified, that operate in a wide variety of industries including government, financial services,
healthcare, manufacturing, IT services, technology, telecom, and consumer products. North America is about 65% of total
revenue. It also has a big client base in Europe.

In terms of the free cash flow generation and DCF, we get the stock's probably worth $180 a share over the long term,
which implies over 90% upside or almost a double if, you know, people stop fearing, um, this AI software, um, contagion.

So ServiceNow's workflow applications are built on its own intelligent platform. The applications built on this platform are
technology, customer and industry, employee, and creator workflows. The products under each of its workflows help
customers connect, automate, and empower work across systems and silos to enable greater outcomes for businesses.
Um, the important thing to understand is that, you know, AI falls into its technology workflows and employee workflows.
So, you know, ServiceNow is an enabler of AI, so it's not something that is ultimately, in our opinion, going to be replaced
by AI.
Um, the Now platform is a single platform with one data model, one architecture enabling speed, productivity, and
innovation. It basically simplifies automation and manual processes for companies. Um, it's frankly a lot easier to use this
model than to create several different AI verticals, uh, that a company can manage itself.

Business is based in Santa Clara, has data centers in North America, South America, Europe, Asia, and Australia. Um, has a
global office base, and while North America counts for 65% of revenue, EMEA is about 25%, Asia-Pacific is around 10%.

Um, so the company has a big AI vision, um, given, you know, from its founder, CEO, it wants to be a single platform that
brings people and agents together to accomplish governed work. It starts with its employee works platform, the
conversational employee front door powered by move works that enables any user to kick off workflows and enterprise
services with access to the web. The company is also determined to automate as many workflows as possible by building
AI specialists that perform business processes and control governance across the enterprise.

So the company has enterprise-wide coverage. The cross-enterprise control plane means that customers don't have to use
ServiceNow's technology to build agents but can still govern them via a central location. So it's essentially going wants to
be an AI, um, management tool. For example, ServiceNow is built across all three hyperscalers so that it can connect
isolated AI investments. Plus, action fabric enables users to leverage their preferred agent development kits and govern
them all using the AI control tower that ServiceNow provides. Areas like ITSM, HR, and security lend themselves to
building in ServiceNow where the data estate is being managed, but not every process needs to be constructed in-house
to be covered.

So, the company has a forward-deployed engineering team. Um, it's announced a project now 2026, giving Accenture
clients access to more than 300 pre-built AI agents, skills, and agentic workflows on the ServiceNow AI platform. The
partnership is targeted, uh, targeting the top of the pyramid use cases, um, that agent skills and agentic workflows on
ServiceNow AI platform have. Um, you know, it's identified problems with multi-million dollar ROIs and stand up sending
up processes in weeks. So basically, you know, we think now is quite interesting as it's trying to become an AI management
platform. It has 8,500 customers, it's been growing revenues at over 30% a year, but the stock's down 60%.

JPMorgan, um, recently had, uh, a conference where, uh, they had several takeaways after speaking with the president,
CPO and COO, Amit Zavery. Um, they said that usage volume and adoption underpin confidence in the company's $1.5
billion AI ACV target. Um, growth engines are fueling a base case, um, for 30 to 32 billion in subscription revenues. So the
company's goals remain intact for revenue growth. Um, its fast-growing CRM business continues to grow. Um, AI confusion
comes from vendor choice and AI solution tangibility. The company acknowledges that customers are exhibiting some
confusion, um, on vendor choice in the tangibility of the AI solutions. ServiceNow explains that its investment to bolster
its AI platform were very strategic and it certainly led to many more at-bats than what we had before. Further, the
company comments that it intentionally gives customers frontier model choice to give them flexibility and the ability to
really do the right thing to run and operate their business as customers look to drive cost savings and efficiencies.
Interestingly, ServiceNow notes that it believes LLMs could potentially become commoditized in the future, likening the
usage of different model versions to silicon chips. Over time, it becomes like chips. You don't really care what chip you're
using in your application. And we've already seen this with Claude, you know, with a number of banks saying, you know,
we want to replace Claude with cheaper models. Even Microsoft is using DeepSeek, which we'll talk about in a moment,
uh, a localized DeepSeek.

Salesforce Share Buybacks - 00:36:32

So, on the following page, on page 17, you can see that Salesforce has bought over $55 billion in stock and the market
hasn't cared. Um, you know, a funny anecdote is if Salesforce were to repurchase shares at the same pace as it did last
quarter, you know, there may not be any shares left outstanding in just 13 quarters. While I say this jokingly, the market
seems not to care at all about the company's share repurchases because it's still terrified of terminal value risk.

Historically, Salesforce didn't prioritize share buybacks. For years, its share count grew because it diluted shareholders to
fund massive stock-based comp and acquisitions like Slack and Tableau. The real change happened over the last three
fiscal years due to immense activist investor pressure. The capital deployment has scaled rapidly. So after no buybacks in
2023, the company bought back $5.7 billion in stock in 2024, 10 billion in 2025. In FY26, which ended Jan 31st, 2026, it
bought back 12.7 billion. Just in March, the company bought back $25 billion in stock.
And year-to-date, it's funded about $27 billion in share repurchases. Salesforce deployed about $28 billion, um, over the
last three years before its recent purchases. And in February of 2026, CEO Marc Benioff announced a massive $50 billion
aggregate share buyback authorization. This program completely alters the company's capital strategy. Instead of slowly
buying back shares using existing business cash flows, Salesforce basically borrowed money, uh, did a $25 billion debt
raise in March, uh, to buy back stocks immediately. So this is called an accelerated share buyback plan, and it immediately
funneled 80% of the $25 billion proceeds into the stock, buying back roughly 103 million shares at once. The rest of the,
uh, 20% of the share buyback plan, uh, took place over April and May.

Um, because it was structurally set up as an accelerated framework with, uh, JPMorgan, BofA, Citibank, Morgan Stanley,
and Santander, the delivery of the shares was front-loaded. Um, but, you know, after the final settlement window, the
company's continued to buy back shares up to the $50 billion authorization, and it's bought back roughly $27 billion worth
of shares, um, year to date, which is very impressive, and we'll see how the market, um, looks at these companies once,
um, you know, some of the software fears are fully understood.

Sats (EchoStar) Puts and SpaceX Unlock - 00:38:54

On the following page, you can see, you know, the aggressive puts buying for, um, for Sats or EchoStar, you know, when
SpaceX IPOed, um, it almost reached 2.8 trillion within a couple of days on June 16th, so it was essentially the fifth most
valuable company in the world on a micro float of about 4%.

Um, the reason why this stock, um, people are, the stock's been so volatile is because, um, you know, the IPO is 4.9%, but,
you know, on August 8th, you know, there's an unlock to 11.8%. Um, there are several unlocks that happen, um, if the share
price is above a certain level. Um, but, you know, by the 180-day point, uh, by November, you will have, you know, over 30%
of the shares, um, unlocked and by, uh, June of of 2027, you know, Musk's 46% shares will also unlock. So the full unlock
will happen by September of 2027. But by the end of this year, you'll have 40% of the shares unlocked, which is roughly nine
times what is available in the float right now. That's why there's been aggressive hedging.

Weekly Market Recap - 00:40:11

Um, so let's take a step back, uh, from, you know, the AGI gold special situation, the FPH real estate special situation, the
three dividend names that we've been adding, Dynex, NLY, and, um, Redwood, and just quickly talk about what
happened in the market last week. Obviously, this will be recorded and you guys can read the PDF as well. But, you know,
a lot can happen in a week. This past week was notable in three ways. First, the US-Iran conflict appears to have reached
the beginning of its end, with both sides expecting to sign a memorandum of understanding to bring about an end to
hostilities and blockades to and begin negotiations. That was slightly derailed, which is why futures are down half a
percent. We obviously wrote this before that, um, after Trump threatened Iran, uh, today, and, uh, Israel continues to
occupy, um, Lebanon.

Second, uh, new Fed Chair Kevin Warsh led his first meeting of the Federal Open Market Committee, um, this week. In a
distant third came a slew of US economic data, some but not all of which came in below expectations. Together, these
three developments further cloud the outlook for the economy over the balance of 2026 and may make the opening
months of Warsh's tenure difficult to navigate.

Warsh's confirmation process was controversial, coming as it did during a period of unprecedented attacks from the
president on the former Fed chair. From the beginning, however, we have not believed his decisions once in the role
would be overly influenced by political considerations. His first press conference, promising major challenges and a return
to first principles, seemed to our to affirm our view. You know, even though Warsh does speak with Besant every week, he
can't afford in the first meeting to look like he's compromised, right? That would be a much worse outcome than what we
saw last week.

Even without the debut of the new Fed chair announcing big changes, as Warsh did, the June FOMC meeting would have
been notable in several ways. First, the FOMC has become more rambunctious lately, with dissents becoming more
routine and individual regional presidents and governors striking out on their own to comment on inflation and the
appropriate path of policy. The new, the Fed's new communication strategy, clearly spearheaded by Warsh himself, is to
say less in statements about its plans going forward or what would make it change its policy rate target. Phrases like, "in
considering the extent and the timing of additional adjustments to the target range of the federal funds rate," which was
in the April statement, are now absent. This probably made it easier to achieve unanimity, as most FOMC members can
agree on statements of fact rather than opinions about the proper path of policy moving forward.

Second, thanks to the conflicts in the Middle East, the US economy has experienced a major economic shift since the
March meeting, which led several forecasts to be revised dramatically, as you can see, with GDP coming down, PC inflation
coming now to the mid-3% range for the year, which we think is is overstated, uh, for a forecast. Um, the economic
forecast revisions once again add a whiff of stagflation with the 2026 GDP revised down to 2.2% as we mentioned, while
expectations for inflation soared to well over 3% for both headline and core PCE prices.

Perhaps most notably, the median forecast for core PC inflation for 2027 also rose by 0.3% to 2.5%, a rate most FOMC
would have deemed to be unacceptably high. The absence of forward guidance for the FOMC statement seemed to give
the members' dot plots greater importance for the markets. The dot plot itself also had a hawkish tilt with just one dot
reflecting a single interest rate cut in 2026, which wasn't Warsh, he declined to submit a forecast, while exactly half the
dots are nine called for one or more hikes. The rest of the committee felt leaving rates on hold would be appropriate.
Compared to the March projection, the FOMC has added 0.375% to its 2026 expectations for the policy rate and 0.5% to its
2027 expectations. That is still not as high as the market expects, but the gap has shrunk considerably.

In terms of the impact of oil supply, it's hard to imagine a geopolitical event that would confuse a central bank more than a
months-long disruption to global oil supply. To wit, the European Central Bank last week elected to raise its policy
benchmark for the first time since 2023. Europe's economy has not grown for two quarters and core CPI inflation on the
continent has only been 2.6% for the past year. In tightening policy, the ECB effectively is attempting to dampen inflation in
parts of the economy it can't control, like housing, to offset high inflation in the part of economy, gasoline prices, that it
cannot. Sorry, it plans to control inflation using housing and and to cap housing prices and rents to offset, you know, this
energy shock.

If you can see US average gasoline and diesel prices, which spiked in 2026 to the same levels that they did during the
beginning of the Russia-Ukraine war. It seems unlikely to us that the Fed will hike rates, especially with oil supply about to
be slowly restored to normal and gasoline prices falling quickly, as the graph above shows. What has become clear,
however, is the surge in energy-driven inflation during the first half of this year is going to prevent the Fed from cutting
rates anytime soon. It's easy to forget now, but rate cuts on a mix of weaker hiring and moderating inflation was the Fed's
plan ahead heading into this year.

Fortunately, the labor market's recent stabilization makes the conversation about cutting rates less pressing for now.
Financial market conditions are loose enough to support tight credit spreads, high equity valuations, and it seems at least
an occasional two-trillion-dollar IPO, uh, which we'll likely see again with OpenAI at 1 trillion and Anthropic at 1 trillion by
December.

The Fed funds rate is only slightly higher than the Cleveland Fed's one-year inflation estimate, and it's well below the
realized headline inflation over the past year. When asked how he felt about the hawkish set of forecasts his colleagues
submitted last week, Chair Warsh noted that projections seem to have been written in pencil and they can be erased in six
weeks. This underscored both the skepticism about economic forecasting in general and a more specific view that
delivering guidance on the economy or rates is unusually challenging right now. We happen to agree with that.

No signs of political interference yet. So judging by Warsh's, um, refusal to provide guidance on markets in the direction of
monetary policy. Heck, it was even hard to get him to deliver a frank assessment of current conditions at his press
conference last week. He probably doesn't want to seem to care much that markets are pricing in earlier hikes following
the Fed's meeting, but they are.

If you look at the following page on 23, you can see the Fed's funds rate with market projections for 2026 and 2027 rising.
Now Warsh's approach for the next six months will be to start from scratch with the Fed's approach to interpreting data,
communicating with the public and managing its balance sheet. He did note that changing the Fed's 2% inflation target will
not be on the table, a move, no doubt designed to solidify the Fed's hard-won credibility on delivering price stability.
The mere belief amongst investors that inflation will eventually get back down to 2% makes delivering on that promise
easier, which is why the Fed has so frequently discussed the anchoring of inflation expectations. On that point, the gap in
yields between the two-year and the 10-year US Treasury note shrank sharply on Wednesday, which typically happens
when monetary policy is tightening. I.e., short-term rates rising faster than long-term rates. This is another sign that
markets do not view politics as a factor in monetary policy so far and are not demanding much of a premium to lend to
the US Treasury for longer periods. So we saw, you know, the Treasury yield curve, the two's 10s cutting from, you know,
basically 70 bips down to 26 bips, you know, from the first quarter of this year.

The risk for any new Fed chair is a significant macro or market shock. Warsh's approach to the job will be to allow markets
to react to events and data rather than to guess how the Fed will react to them. That is a worthy and admirable goal, but
should a market crisis arise, even one that is not of the Fed's own making, investors will want to understand how and how
much the Fed is committed to help in the form of cuts and liquidity injections.

We do not expect a moment of truth in which the Fed decisively either tightens or loosens monetary policy to pop up this
year, but we can say we can easily imagine scenarios in which inflation roars higher due to resumption of Middle East
hostilities or the job market starts to soften with consumers getting squeezed.

Perhaps the Fed has over-communicated and over-promised during the last three Fed chairs' tenures and Warsh's
chairmanship will be a needed correction from the central bank's tendency to talk more, write more, and do more. But the
transition between the Bernanke-Yellen-Powell approach and the new one could be rocky, at least at the outset.

Iran Peace Negotiation Timeline and Breakdown - 00:49:50

On the following page 24, we discussed the five-day Iran peace negotiation timeline with detailed 14-point description of
the memorandum of understanding or MOU, and then today Iran walking out of negotiations after new threats. So
basically on Monday, um, we had a MOU to be signed. Um, you know, Trump and Vance signed the MOU, so did Iranian
parliamentary speakers. Um, the deal provided for immediate opening of the Strait of Hormuz and lifting of the US
blockade. Details of the agreement were then released. On Wednesday, um, and there's discussion of relief of frozen funds,
plan to keep, uh, US military force posture during the next phase of negotiations over 60 days, and an agreement to
contemplate a reduction of military forces upon the agreement of a final deal. Brent basically fell below $80 for the first
time. You know, WTI fell below $78. It's effectively, you know, modestly higher, um, overnight.

On Tuesday, there's an Iran MOU draft update where Iran said they don't, they agree not to develop or acquire nuclear
weapons. Um, US and Iran agreed to halt hostilities across the entire region, including Lebanon, which Israel has not done,
which is why Iran has walked out. Iran also said they would guarantee free and safe commercial shipping through the
Strait for 60 days. The US was then to release frozen, uh, Iranian assets once the agreement took effect.

So, both sides agreed to address Iran's uranium style pile in future negotiations, so that was not agreed to. Um, and on
Wednesday, there was a flip-flop with Trump, uh, saying that, um, you know, he would sign the deal by Friday. Um, and
then he said that Iran's MOU is not final. If I don't like the agreement, we'll be going back to dropping bombs. If Iran isn't
behaving, they will get hit again. Uh, Trump said that the US is not putting up money for Iran, but this $300 billion has to
come from somewhere. No one right now knows where it's coming from, which is hilarious. Um, Trump said that they will
only lift Iran oil sanctions if they do things right, which makes sense, actually. Europe is doing some things badly. Um,
Trump also said that the Lebanese president is coming to the US in the next week or two and that Lebanon had been
treated worse than anyone.

At the G7 meeting on Wednesday, Trump, uh, on the Iran deal said that maritime traffic in the Strait has increased
substantially, which it did temporarily, but now it's back down. Um, it's really the Iranian vessels that, uh, escaped, uh, on
their way to China, but non-Iranian vessels pretty much had been blocked by the end of the week.

Um, as you can see now, the US Nasdaq is actually down 80 bips because of, uh, fears about the war restarting. Small caps
are now down about 80 bips and the SPX is down about 55 bips, um, which shows that the concern around escalation, uh,
and another energy shock is only accelerating.
Quickly looking to see, apparently a US diplomat told Barak Ravid that the talks with Iranians started on Sunday morning,
um, and were going non-stop. Um, one of the issues was deconfliction mechanisms in Lebanon and enforcing the
ceasefire. The US diplomat added that the US, Iran, uh, mediators discussed the Strait of Hormuz and recent Iranian
statements about allegedly closing it. They wanted to make sure it was stayed open. Um, the US diplomat said that the
parties had productive talks, um, but then, um, while they were pleased about how talks went today, um, Iran then left
after continued bombing of Lebanon.

Uh, so that's, that's basically, I think what happened, and then Trump effectively threatened Iran by saying that its leaders
wouldn't be allowed to return home without a peace treaty. Um, and then the the Pakistan, uh, Prime Minister was frankly
shocked, um, when Iran left because he thought this was a done deal. Um, you know, the 14 points of the Iran-US MOU
are on page 25, including an end to conflict on all fronts, which has not happened, respect for internal affairs, an
extendable 60-day timeline, US to end the blockade, uh, Iran to end its blockade, money for Iran's reconstruction, um,
sanctions to end on point seven, no new nuclear weapons for point eight, points nine and 10, basically a status quo where,
um, US and Iran agree that the nuclear program will be, uh, halted. Um, I'll point 11, Iran's frozen funds to be released.
Points 12 to 14, around monitoring and final negotiations. You know, the US is, you know, is being very careful not to, uh,
agree to remove sanctions or give funds until final negotiations are complete.

So that was signed on Wednesday and then on Thursday, um, Israel started to, uh, aggressively bomb Lebanon and then on
Friday, these attacks delayed negotiations. Um, Vance first canceled his Switzerland trip, then restarted it. Um, and on an
interesting perspective, 11 tankers holding about 20 million barrels actually left the Iranian port of Chabahar on the Gulf of
Oman this week. So Iran was actually able to send its oil, but shortly after it sent its oil because of the continued bombing,
it actually stopped allowing, uh, non-Iranian vessels to leave. Um, then on Sunday, Iran walked out of talks, um, after
Trump said that Iran must immediately stop their highly paid proxies in Lebanon from causing trouble. If they don't, we
will hit Iran very hard again, just like we did last week, only harder.

Um, so whether that's right or wrong, you normally don't say those types of things like during an actual peace meeting,
um, which was not taken well. Um, so now basically Iran, and that's actually never been done in history, you know, during
a peace, a live peace negotiation threatening another country with bombs. Um, so Iran now said that they will not return
until they receive an apology, which they probably won't, and Lebanon attacks stop, which they probably won't. So Iran
basically, um, has said it won't return to Switzerland, and I don't expect they'll return for for Switzerland for several days
now. Um, and there's really no, um, end to the conflict in sight. Bill Ackman, uh, smartass, basically said he doesn't
understand who is going to invest in a $300 billion reconstruction fund for Iran. Obviously, the only country that has those
types of funds would be the US, uh, and some Middle Eastern nations. But the funny thing is, people say the US is not
going to fund it. Who the fuck do you think is going to fund it?

Saudi Arabia, Qatar, all these Middle Eastern nations have promised, you know, over a trillion dollars, uh, in US
infrastructure, right? If the US is not going to fund it, they're just going to take, they're just going to take back what they
were going to spend on the US and fund it. There's no one else in the world that's going to fund Iran's reconstruction. So
it's either going to come from the US or it's going to come from funds that were promised to the US from the Middle East
now that the Middle East has had tens of billions of infrastructure projects destroyed. And for anyone who doesn't
understand that, you know, you read, you need to gather some brain cells. It's either going to come from the US, US
companies, or it's going to come from the Middle East that is going to take away funds that were going to be spent on the
US, uh, that they can't afford to now spend on the US because they have to rebuild themselves. Um, and that's why we
don't trust the government because they're completely full of shit and can't explain that simple concept to the public.

FOMC Meeting Details and Warsh's Approach - 00:58:00

Um, so the very specific details on the FOMC meeting, almost excruciating detail on Warsh, uh, and the minutes, uh, you
can read on page 31. I'm not going to spend a lot of time going through this because we did summarize it earlier in this
piece. Uh, so on page 32, Warsh's, uh, 2:30 p.m., uh, speech, um, and page 33 is discussed, and then you can see FOMC
members see inflation risk, they're less worried about jobs in the white and blue lines on page 33.
Um, and then on page 34, you can see a graphical representation of real GDP forecast, unemployment rate forecast, core
PC inflation forecast, um, in the March SEP and then the June SEP. So you can see a spike in inflation expectations, a drop in
GDP, and then an expected Fed funds rate, uh, on page 34. And then you can see the yield curve flattening almost real-
time in the in the meeting, and then a, um, 15 basis point spike in the two-year yield at the bottom of page 35, and then,
you know, the nine FOMC members that called out rate hikes, and the new FOMC minutes that were cut back about 70%
on page 36.

On page 37, you can see that the Bank of Japan is gradually raising, uh, its interest rates, uh, up to 1%, which is the highest
since 1995. ECB also hiked in June and said that hikes are not over, especially if there's no peace. The ECB could actually
hike in July, which in my opinion would send Europe into a recession. And that's not being priced in by the market at all at
the moment, which could mean that European stocks could be a short and European credit could be a short if the ECB
hikes again in July.

We are going to see PCE on June 25th. Um, with the May PPI and CPI in hand, forecasters expect core PCE to print around
0.35% in May. That would raise the year-over-year rate to 3.4%. The six-month annualized PCE would climb to 4.1%, which
is the highest since June of 2023, which is why the Fed is so scared. Both measures were below 3% in the year-earlier
period. And basically this war has crushed crypto, it's crushed gold, and it's crushed real estate, uh, because there's no way
the Fed can actually cut interest rates this year. As I'm speaking, the US Nasdaq is now down almost 1%. The small-cap
Russell is down about 80 bips and the S&P 500 is down about 60 bips.

Data Center Delays - 01:00:33

The other important topic that we, uh, discussed last week and many people are discussing is data center delays. Uh, so
Jefferies and JPMorgan wrote some pretty detailed pieces about how they think, you know, 70% of data centers that were
supposed to be built by 2027 haven't even struck ground. Um, semi-analysis challenged this to a bit, but they didn't have
like a really good conclusion. I think they're basically saying that it's not 70%, it's under 50%, but there's still a huge delay in
data centers because of power bottlenecks, memory bottlenecks.

Um, the $700 billion AI infrastructure buildout hit a wall that capital alone cannot fix. And Jefferies laid out why. If you can
see in the table below, um, the data center pipeline chart shows an announced capacity surge surging from 6.9 gigawatts
of data centers in 2025 to 24.1 gigawatts in 2026, peaks at 44 gigawatts in 2027 and sits at 40 gigawatts in 2028. This
pipeline looks like an unstoppable infrastructure build out, except for the fact that most of the announced data center
builds have not even been started yet.

Jefferies basically says that about 50% of 2026 data centers and 80% of 2027 and 28 data centers do not appear to have
even started. JPMorgan independently confirmed it, finding that over 60% of data center capacity planned for 2027 has
not broken ground, with another 7% already flagged as delayed. Site Light Climate put the raw numbers behind it. Of the
roughly 12 to 16 gigawatts of US capacity announced for 2026, only about 5 gigawatts is actually under active
consideration, meaning the rest is still paper.

The reason is that so much of the announced pipeline is stalled is not capital, but just physics. High-voltage transformer
lead times that ran 12 to 18 months in 2022 have extended to 30 to 36 months. There are simply not enough transformers
to build the power necessary for these data centers to be completed. Some of the voltage transformer lead times are up
to five years, a catastrophic mismatch for data centers that need to deploy them within 18 months to match AI demand.
Grid interconnection queues in high demand areas like Texas, Virginia, and the Carolinas have grown to three to five years,
meaning that projects that broke ground in late 2024 and expected to be online by mid-2026 are finishing construction and
finding that they can't physically receive power or put GPUs to work. Capacity market clearing prices in these regions
reflect that scarcity, hitting $329 per megawatt day in 2026, up from $28.92 the year before, a 1,037% increase in a single
year. Power generation cancellation data makes it significantly worse. The second chart on page 40 shows postponed
power generation projects across every major category: solar, natural gas, batteries, onshore wind, and offshore wind, all
trending sharply higher from April 2023 through April 2026. Battery cancellations in Texas's ERCOT, which Jefferies
specifically flagged as a key forward indicator, are accelerating precisely when they should be expanding to support new
data center demand. The conclusion is that headline pipeline figures materially overstate the capacity that will actually
come online. Solar and battery project attrition is accelerating just as data center demand firms and the supply
environment is getting tighter. What this means for the AI infrastructure trade is direct and consequential. When 50% of
the 2026 capacity and 80% of 2027 capacity is on paper rather than under construction, the AI compute shortage does not
ease on the timeline the market is currently pricing, but rather actually extends. Every gigawatt that slips from 2026 to
2027 or 2028 is a gigawatt of compute that hyperscalers, model labs, and enterprise AI teams cannot access, which means
they will pay a premium for whatever capacity actually exists. And the companies already that already have operational
data centers and secured power today are not in a competitive market. They're in a market where they will receive a
premium. They're in a seller's market with no meaningful competition for longer than consensus currently believes. So this
is an environment where companies like DGX, uh, where CoreWeave, um, where Nebius, where all these names will, uh,
have higher lease rates until data centers can actually be built. So despite, you know, Chanos being negative on the space,
um, it still has another year or two, I think, before, at least a year, before this bubble really blows up given all the delays
that we're seeing in the data center build out.

Now, semi-analysis, you know, uh, Nutson, Barkin, and Onteveros, uh, wrote a piece on June 18th that basically said that,
hey, while the data center, uh, capacity has been delayed, uh, it hasn't been delayed as much as, um, these banks are
saying. You know, the first rumors of big delays came on April 1st in Bloomberg's piece, America's AI build out hinges on
Chinese electrical parts, which framed the 2026 capacity slowdown as a consequence of fragile, China-dependent
equipment supply chain. Okay, so that was actually partially true because a lot of the voltage transformers, etc., come
from Asia. Bloomberg, you know, didn't lead with that framing, but within days, TechRadar, Tom's Hardware, the Register,
and other news outlets ran sharper, more clickbait versions claiming half of data centers were canceled, and that's the
version now circulating.

Basically, uh, semi-analysis said that they were first and, um, it's really a bunch of, you know, kind of a, it's kind of a bullshit
piece to get you to buy their database. Um, but what they're saying is they're the first to to note that colocation delays and
data and hyperscalar delays were happening. Um, but the delays are not as bad as JPMorgan and Jefferies say. But the
conclusion is really that data centers are highly delayed. Whether it's 40% or 70%, um, there's going to be demand for
compute that far exceeds supply. Um, you know, in the near term.

Microsoft, OpenAI, and the Shift to Usage-Based AI Pricing - 01:02:59

In the longer term, we're seeing, you know, that that, um, corporations are starting to balk at AI pricing, and even
companies like Microsoft are now including DeepSeek native models as a way to, um, lower their AI spend. So Microsoft
plans to move Copilot to usage-based pricing with open-source AI model DeepSeek instead of OpenAI or in addition to
OpenAI due to high cost of implementation. Even though this represents less than 5% of pro-forma revenue for OpenAI,
the risk to OpenAI ahead of its IPO is that others could follow suit, creating a bigger issue.

Microsoft CEO Satya Nadella's shift to offer DeepSeek version 4 alongside a move to usage-based token and pricing for
Copilot's co-work is an absolute watershed moment for the AI industry. The era of all-you-can-eat flat-rate corporate
subscriptions is ending because agentic AI, AI that continuously plans, codes, and loops in the background is dramatically
more expensive to run than simple chatbots. The financial and strategic ripple effects across tech hardware and venture
capital break down as follows:

AI models are cost-prohibitive. So OpenAI's ChatGPT 5.5 and Anthropic's Claude 4.8 have priced themselves into a
corner. Reportedly, DeepSeek's version 4 costs roughly one-third as much as Claude for inputs and one-seventh as much
for outputs, then taking away pricing leverage. Microsoft's, uh, Microsoft is OpenAI's biggest backer, but Nadella is
explicitly executing a multimodal platform strategy. He noted on X last week that we do not want a world where
enterprises seed all value to a small number of models. By introducing a hyper-efficient open-source competitor,
Microsoft is stripping OpenAI of its premium pricing leverage.

Now, what could this mean for hardware? Um, the shift to open-source, highly optimized mixture of experts or MOE
architectures like DeepSeek means that enterprise software can achieve the same results using significantly fewer
compute cycles. If the industry shifts from bloated massive frontier models to lightweight, fine-tuned open-source models,
the frantic exponential panic buying of premium AI chips could cool down longer term.
On the bull side, um, the AI is moving to a usage-based enterprise billing, which means that there will be a lot more AI
consumed. This transforms AI into a metered utility like electricity or AWS cloud storage, and enterprise enterprises
optimize their their cost via cheaper models, total volume, and actual utilization of AI agents will skyrocket. The sheer scale
of agentic workflows running continuously will still require an immense, unyielding baseline of infrastructure. This
development explains exactly why the public market has started applying heavy discounts to proxy investment vehicles
holding pre-IPO tech blocks. Closed and venture vehicles like DXYZ or VCX that we are heavily short, uh, or even Parallax
PWRL, all have concentrated secondary allocations in OpenAI and Anthropic. Both of which are currently in the
confidential stages of preparing for trillion-dollar IPOs. With Microsoft proving that it will happily swap out open-source
models to preserve its own corporate margins, the public markets are realizing that the long-term enterprise software
moats of OpenAI and Anthropic are far less secure than originally hyped.

There's also a political issue. So this move introduces a massive geopolitical paradox. The Trump administration has been
actively tightening restrictions on advanced AI models leaking out of the United States. Now, Microsoft, one of America's
most financial tech institutions, is adopting a model born out of a Chinese lab, simply because the raw economics of
American frontier models are unsustainable. To protect itself from political blowback, Microsoft has heavily modified the
model to make its make it its own, layered it is with Western safety bias overrides, and restricted it entirely to self-hosted
Azure environments.

You know, on a similar note, JPMorgan also blocked, uh, using Claude in Hong Kong because of its cost, but also because
of some compliance issues. On the following page 44, you can see that, you know, screenshots of the JPMorgan um,
execution, and then on page 44, you can also see the AI trade is extremely crowded, you know, with, um, according to
BofA, you know, 80% of investors saying that AI trade is too crowded versus, you know, roughly 72% in May and, uh, only
25% in April of 2026.

If you look at the options flow of Micron, Marvell, Intel, Dell, or Oracle, um, on Friday on page 45, you know, they were at
unbelievable highs. Um, you know, AI was almost every single, uh, option, was the highest, uh, volume across those five
five companies. Um, and one interesting thing is that despite the the boom in AI, just like we saw the boom in internet, um,
you know, biggest companies that should benefit from AI are actually underperforming. So Microsoft's stock is down over
20%, you know, from its highs and now Michigan pension fund is suing them because they said that AI cloud computing
profitability would be higher than it really is. Um, when, you know, Microsoft obviously has an, I believe that this lawsuit,
you know, they won't win the lawsuit, but it just shows how frustrated investors are. You guys can read the piece.

Um, and what we've noticed also is that private credit withdrawals continued. We had another, we've had several billion of,
um, withdrawals from interval funds. Thoma Bravo took a $5.1 billion write-down, and Blackstone took another $2 billion
write-down on, um, along with the other lenders because there's 3 billion in debt, um, and the private credit loan take-
back paper is now 40 cents on the dollar. So there's roughly an 85% write-down on that structure. And it kind of just shows
you that the private credit contagion is frankly not over. The market's been ignoring it, but eventually defaults will rise.
They're about 6% according to, um, to Fitch right now.

So, you know, we quickly went through that 47 pages. Um, again, to wrap it up, you know, we think that Warsh, um, will
reveal more of his true colors six weeks from now depending on the Iran war peace treaty. PCE will be very important next
week. In terms of earnings, Micron will be the most important earnings next week. Um, in terms of special situations, we
talked about the Honeywell spin-off, HONA. We talked about, um, Caesars potentially selling its Atlantic City assets. We
talked about Roku being bought by Fox A. We talked about Netflix not being interested in Roku or Lionsgate. We talked
about, um, Warner Brothers. We talked about, um, Two Harbors, uh, in the merger section. We talked about bare
positioning in gold being too high and adding gold exposure along with the Alamos Gold, uh, which we'll double again. So
we doubled from five bips to 10 bips. We'll likely double it via US shares on Monday, especially if they're down with this, uh,
this market sell-off due to the, um, due to Iran leaving Switzerland today.

Um, then we talked about, um, the real estate special situation, FPH, potentially being a triple, uh, because of its prime
quality real estate near Orange County, at Great Park neighborhoods. Um, we also talked about adding to high dividend
names like Dynex, NLY, and Redwood with 14 to 16% dividends on this, uh, rate, uh, rate hike, uh, fear. And, um, we
talked about, uh, ServiceNow being quite cheap, uh, and discussed briefly our DCF model on it and the company's
business model. Then we talked about how the market has been ignoring, uh, Salesforce's 55 billion share buybacks so
far over the last three years and the accelerated 25 billion share buyback in March. Then we talked about, uh, SpaceX's
unlock and, you know, uh, how to hedge Sats and, uh, at the money put buying in Sats from probably SpaceX institutional
holders.

Then we talked about the FOMC, the rate market, the Iran peace deal. Um, we talked about, um, the FOMC and, uh, the
yield curve flattening. Then we talked about the Bank of Japan hike and the ECB hike, and then we talked about PCE and
data center delays. So pretty detailed call. Um, we'll quickly go into Q the Q&A tab. We don't see any questions, so we'll
end this call for today. Um, so we had, you know, several recommendations in this piece from gold special situations and
ETFs to software, to real estate, um, land-based, uh, holdings that are interesting. Uh, and we discussed why futures are
down 1% now with the Nasdaq. So I hope you guys have a blessed trading week. There will likely be, uh, names that we
can buy, uh, next week, especially if oil were to spike further. Brent's already at 81.

Um, as you guys know, we bought oil last week, um, small amount of oil, um, given, you know, as a hedge, given we were,
uh, worried of this, uh, peace deal happening, and it turns out we were right. Wish we bought more oil. Um, but because
the the peace deal is, you know, is basically 75% done, um, the market's really not, um, pushing oil much higher. If it takes
another few weeks for this peace deal to be signed, you could see oil going higher from here. Uh, so the risk-reward of oil
when we bought it in the mid-70s, I thought were pretty decent. Um, so we'll close the call now. The recording will be up in
a few minutes. You have the PDF, you have Joel Greenblatt's class on special situations, which we thought was very well
done. You have the ServiceNow model, you have the AGI gold model, you have the FPH real estate write-up, uh, and you
have the Standard Chartered piece on the Warsh FOMC meeting, which was a good read, HSBC on broadening market
performance, DB on the Anthropic Mythos model, and the BlackRock earnings piece. So again, wish you guys a good
week and we'll talk to you soon.

---
_Readable notes for personal study. Full premium transcript & report retained as PDFs in this folder. Not investment advice._