Max

32% profit/year since 2020
🇩🇪 Top 10 eToro Popular Investor
$5m+ AUM | Not financial advice, DYOR

[Max @mrmagoon]

💥 𝐌𝐚𝐫𝐤𝐞𝐭 𝐌𝐨𝐯𝐞𝐬 𝐓𝐞𝐥𝐞𝐠𝐫𝐚𝐦 𝐆𝐫𝐨𝐮𝐩

I’m putting together a small private group of ~50 high-quality eToro investors.

If you want in, I’ll drop a link to the waitlist below:

https://t.co/UVrr76LDwo

Once we reach 50 waitlist entries, it will be closed and the Telegram group will be opened.

Let’s keep building growth.

[Max @mrmagoon]

Meta just shipped the most ambitious consumer AI product of the year

And the man behind it explained why in a piece that barely has a single number in it.

Alexandr Wang, Meta's Chief AI Officer, published "Why I'm Building Muse" this week.

I expected specs, benchmarks, a roadmap.

Instead I got a short, almost emotional manifesto about human wishes.

His starting point is simple.
Most people want things badly.
More time with family.
Better health.
Their own business.

But most of those wishes never turn into action.

Life is full of forms, gatekeepers, calendars and "who do I even call for this?"

Most dreams die of hesitation, long before they die of failure.

Muse is his answer.
An AI agent that works like a personal manager.
You bring the wish.
It builds the plan, handles the logistics and does the talking.

Now let's add the numbers the manifesto leaves out.

Muse launched on September 8, first in the US, on iOS, Android and web as a free and paid version. Every agent runs in its own cloud machine with a visible browser.

Within roughly 10 days it became the number one free iPhone app in the US.

At Meta Connect on September 24 came the next layer. A realtime avatar with smart glasses support. Desktop control on the Mac and its own email address for the agent.

Retail partners like Walmart, Best Buy, Wayfair, Expedia and Instacart. Checkout through Stripe, Shop Pay and PayPal. More than 1,500 developer connector submissions in under a week.

Zuckerberg says Muse stays free for lots of usage. Meta wants to earn a small fee on the transactions instead. Thats wild! Think about that for a second!

Meta spent two decades monetizing your attention. Muse is the first serious attempt to monetize your intentions.

That's a massssssive difference…

Ads sell the chance that you might buy. An agent that books, orders and pays sits right at the moment you actually do.

The companies closest to that moment:

The platform
$META

Checkout and payments
$SHOP
$PYPL

Commerce partners
$WMT
$BBY
$W
$EXPE
$CART

The one keeping its door closed
$AMZN

Amazon currently blocks Muse from ordering on its site. And honestly? That makes total sense to me.

If an agent chooses where you buy, the storefront loses its power.

Now the other side.

Many of the early savings success stories came from Meta employees, and none were independently verified.

A security researcher found a zero day in the Mac app that allowed session takeover. It's patched, but he disputes Meta's low risk rating.
Agents that act on your behalf can also break things. Deleted files, cancelled accounts, wrong orders.

And privacy questions around an agent that knows your wishes, your cards and your inbox are only getting started.

So where does this leave me?

Compute was the first bottleneck.
Memory and power came next.
The next constraint might be… trust!

Whoever gets people to hand over their wishes and their wallet to an agent owns the most valuable interface of the next decade.

Muse is the first real test of whether people will do it.

Would you let an AI agent spend your money?

—

I’m running a small private Telegram group of high-quality eToro investors.

If you want to join, visit my X account (link on eToro profile). I posted a link to the waitlist for the group there. Once you applied comment on this post.

Let’s keep building growth,
Max

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

The Fed just raised rates for the first time since July 2023

25 basis points to 3.75% to 4%, on a unanimous 12 to 0 vote…

And I keep coming back to one uncomfortable question.

The whole playbook rests on a single assumption.
Higher rates make money more expensive.
Expensive money cools demand and investment.
Cooler demand brings inflation down.

But what happens when the biggest source of demand in the economy barely cares about the price of money?

Amazon, Microsoft, Alphabet and Meta plan roughly $725B in capex for 2026. That's up about 77% from around $410B in 2025.

Meta raised its guidance twice this year. Evercore and Bank of America already model more than $1 trillion for 2027.

If you believe you're racing for superintelligence, the payoff looks close to infinite. A few extra basis points won't stop you from ordering the next gigawatt of compute.

Interest is a cost too!

It sits inside every data center loan, every power plant, every grid upgrade, every inventory line.

Economists even have a name for it: the cost channel of monetary policy.
When rates rise and demand doesn't fall, part of that higher cost can end up in prices.

Follow that logic one step further and it gets uncomfortable.
Higher rates feed higher costs.
Higher costs feed stickier inflation.
Stickier inflation pushes the Fed to hike again…

The new dot plot already points to 4.1% to 4.4% by year end. And the Fed itself doesn't expect to hit its inflation target until 2029.

So am I right?

Honestly, I think I'm only partly right.

Rates still bite hard across the rest of the economy.
Mortgages, small businesses, consumer credit and commercial real estate feel every single hike.

A big chunk of today's inflation also comes from energy.
Months of war in the Middle East pushed energy prices up, and no rate hike on earth drills a new well.

And the AI race isn't fully rate proof either. Big Tech has been issuing tens of billions in bonds to fund the buildout.

Amazon alone raised $24.9B in July. Every hike makes the next bond more expensive.

Where rates barely matter, because cash flow funds most of the race:
$AMZN
$MSFT
$GOOGL
$META

Where financing costs matter a lot more:
$ORCL

The Fed's tool is blunt

It hits the rate sensitive economy first and the AI buildout last.
That means the Fed may have to squeeze housing and Main Street even harder to offset a capex boom it can barely touch.

Whether that's a mistake comes down to one question.
Does all this AI spending create enough productivity to pay for itself?
If yes, today's inflation is the entry fee for a massive supply boom later.
If no, we're financing the buildout at 4% and rising.

What am I doing about it? I stay invested in the bottlenecks: compute, memory and power, because that's where the capex lands no matter where rates go.

But I'm watching balance sheets more closely than ever…

So tell me. Did the Fed just make a mistake, or am I overthinking this?

–

I’m running a small private Telegram group of high-quality eToro investors.

If you want to join, visit my X account. I posted a link to the waitlist for the group there. Once you applied comment on this post.

Let’s keep building growth,
Max

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

On Wednesday last week the Fed delivered its first rate hike in three years.

On the next day, CoreWeave asked the market for another $3 billion.

The stock fell to $79.88, its lowest close in over a month. Then Cathie Wood's ARK stepped in and bought roughly $19 million of it. By Friday $CRWV closed at $81.36.

Let me unpack what actually happened…

What CoreWeave announced:

$3.0B of convertible senior notes due April 2033, with an option for $500M more. An indicated coupon of 2.375% to 2.875%. A conversion premium of 22.5% to 27.5%. Plus an at the market program to sell up to 35 million Class A shares over time.

Translation: a convertible is a loan that can turn into shares if the stock rises enough. Because lenders get that upside, they accept a tiny coupon. CoreWeave borrows billions at under 3% in a world where the US government pays 5% on its own 10 year bonds.

That is the whole trick. Part of the proceeds even buys capped calls, basically insurance that limits how much dilution existing shareholders eat if conversion ever happens.

Now the balance sheet this lands on:

Long term debt around $27.6B. Total liabilities around $72B, resting on roughly $5B of equity. A Q2 net loss of $626M. Capex guidance of $35B to $39B for this year alone.

And the demand side of the same company:

Q2 revenue $2.575B, up 112% YoY. A backlog around $104B as of June 30. Short term AI compute contracts reportedly pricing near $40M per megawatt of annualized revenue.

So which is it, red flag or rocket fuel?

Here is the test I use, the same one from the $AAOI episode in August: what does the new money build?

Dilution that plugs operating losses in a shrinking business is a warning. Dilution and debt that build capacity customers have already signed for is closer to a mortgage on a rented out building.

CoreWeave's raise sits mostly in the second category. $104B of backlog wants delivery, and delivery costs capex up front.

The Fed just started hiking. The 10 year sits at 5%. Every neocloud business model is a spread: borrow at one rate, rent GPUs out at another. The hike pushes up the borrowing side. Competition sets the renting side. The spread is the business. Its all connected.

These notes mature in 2033, which is so soon! The GPUs they finance will be several generations old by then. A mortgage on a building assumes the building holds its value. Nobody knows yet what a 2026 GPU earns in 2033.

That is why the same week produced a wave of selling and a Cathie Wood purchase. Both sides are reacting to real numbers. They just weight them differently.

I own the layers that get paid cash on delivery, whoever wins the rental war: $MU $SNDK

And I am looking at the power layer that gets paid before a single token is generated: $VRT $ETN

The landlord layer, $CRWV and $NBIS, I watch closely. Respect for the growth. Respect for the leverage too

Open questions I take quite serious: If GPU rental prices fall before 2033, the spread compresses from both sides at once. Customer concentration is real across every neocloud. A capital markets freeze would hit debt funded builders first, and that risk rises in a hiking cycle. Definitely something to think about.

This cycle keeps finding new bottlenecks. First compute. Then memory. Then power. The Fed may have just created the next one: capital itself. When money costs more, business models built on cheap money get repriced, while business models built on invoices keep collecting.

Watch who can still borrow at under 3% a year from now. That list will tell you who the market really trusts.

—

I’m running a small private Telegram group of high-quality eToro investors.

If you want to join, visit my X account. I posted a link to the waitlist for the group there. Once you applied comment on this post.

Let’s keep building growth,
Max

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

This morning, before the open, $BE joins the S&P 500 and $SNDK joins the S&P 100…

For two weeks I kept seeing the same question under both tickers: do you buy before the index funds are forced to buy, or after?

So I went looking for what the data actually says about the famous index effect.

In the 1990s this trade was a money printer. A stock added to the $SPX500 gained a median 8.3% between announcement day and effective day. That is S&P's own research.

Then the edge started dying though

2000 to 2010: additions gained a median 3.6% aaand 2011 to 2021: essentially zero. Statistically nothing.

Harvard research by Greenwood and Sammon found the same decay from another angle. The total addition effect went from +7.6% in the 1990s to +0.8% in the 2010s. Indistinguishable from noise.

What killed it? Sunlight. Index changes are announced weeks in advance, everyone sees the flows coming, and the index funds themselves got smarter about trading around the event.

Now look at the live example

$BE jumped 9% on announcement day, September 4. On Friday it closed at $265.63, down 5.4% on the day. Still up roughly 205% in 2026.

$SNDK rose 11% on Friday alone to $1,791.82. Up more than 650% this year, the best performer in the entire S&P 500. And still about 24% below its June peak.

The pattern is old and boring: the pop lives at the announcement. By effective day, most of the flow is already priced.

What inclusion still gives you is quieter and slower. A permanent passive bid. Deeper liquidity. More institutional eyes on every report.

What it cannot do is run the business…

For $BE the question remains whether revenue really doubles this year as guided, $3.9B to $4.2B. For $SNDK it remains NAND pricing power through 2027 abd if HBF takes off.

Risks, honestly: Inclusion flows can whip prices around for a few days. Short term noise cuts both ways. Stocks up 200 to 650% YTD can correct hard on nothing at all.

Zoom out: an index seat is a graduation ceremony. The studying happened in the years before, and the exams that matter come after.

Today the ceremony. Tomorrow, back to earnings.

I’m running a small private Telegram group of high-quality eToro investors.

If you want to join, visit my X account. I posted a link to the waitlist for the group there. Once you applied comment on this post.

Let’s keep building growth,
Max

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

$BE joins the S&P 500 before the open on Monday.

The stock jumped 9% on the announcement and sits at around $265, up about 169% in 2026.

The business is real. Q2 revenue $1,065M, up 165% YoY. First billion dollar quarter ever. A $25B partnership with Brookfield to power AI data centers.

Fuel cells solve a real bottleneck: data centers, like those of $NBIS or $CRWV cannot wait years for grid connections.

all three of these major AI infrastructure "Neocloud" operators contract with Bloom Energy..

Nebius signed a massive 10-year, $2.6 Billion master agreement with Bloom Energy in May 2026.

CoreWeave was one of Bloom's earliest high-profile validation partners in the AI sector, signing an on-site power generation partnership in July 2024.

Bloom solves the energy bottleneck by bypassing the grid entirely. Companies who can’t wait for local grid upgrades buy Bloom’s fuel cells to generate on-site power right next to their data centers off-grid.

Basically the opposite strategy of $IREN , which beat the bottleneck by acting early, acquiring land in rural or energy-abundant areas (like Childress, Texas), and natively securing 2.9 to 5 gigawatts of grid-connected power capacity before anyone else can claim it.

Now the uncomfortable number…

Retail chat rooms calls Bloom a $1,000 stock.

The average target across 29 analysts? $276.05.
That is last weeks Thursday's closing price. The street thinks the next 12 months are already paid for.

Right theme and dangerous entry can be the same stock at the same time.

Index inclusion buying is a one time flow. Power demand is a decade long story.

Are you buying?

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Piers Morgan @piersmorgan]

Incredible. Congrats @elonmusk - this may be your finest creation. 👏

[Ren @ren_stocks]

I'm long $SNDK

On 16 June I published my first deep dive on SanDisk, by far one of the best performers in my portfolio

Three months later the thesis and the bull, base and bear numbers are still on track

Take a closer look and build your own conviction https://t.co/YmLkG7lpHq

[Max @mrmagoon]

In February, Nebius reported quarterly revenue of $227.7M

Consensus wanted about $244M. A 7% miss…

Reddit sentiment on $NBIS collapsed from 95 to 35 in days. Panic selling everywhere.

But: the company said its AI cloud capacity was fully sold out.

Read that again. Sold out!

There are two kinds of revenue misses, and they mean opposite things.

A demand miss: customers didn't show up. The story is weakening.

A supply miss: customers showed up faster than the company could build. Every rack gets
rented the moment it exists.

In a bottleneck industry, the second kind is the thesis working, printed in an ugly font.

What followed:

Q2 2026 revenue $582M, up 454% YoY. A $MSFT contract worth up to $19.4B through 2031.

The $META deal, which started at $3B, expanded to roughly $27B in March. Nebius around $224 as of Friday, up roughly 167% in 2026.

The February panic sellers sold the bottom of a supply constraint.

And honestly? I understand why. A red print and a falling chart feel identical whether the cause is demand or supply. Your broker app doesn't label the difference.

So this is the question I now ask before reacting to any miss in AI infrastructure:

Did revenue miss because capacity sat empty?

Or because capacity ran out?

The risks have not disappeared, to be clear:

Nebius carries heavy capex and rising debt to build that capacity. Customer concentration is real when two hyperscalers dominate the backlog. A supply constrained thesis dies the day supply catches up.

The whole AI supply chain is currently a chain of sold out signs. Memory, power, optics, GPU racks.

In that world, quarterly revenue tells you what could be delivered and the backlog tells you what was demanded. A mirror into the past and future.

—

I’m running a small private Telegram group of high-quality eToro investors.

If you want to join, visit my X account (link on eToro profile). I posted a link to the waitlist for the group there.

If you enjoyed "Market Moves", make sure to leave a like and follow me. I'll be back tomorrow.

Let’s keep building growth,
Max

Copy Trading is not investment advice | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

On September 4, a poll asked retail traders how many rate hikes the Fed would deliver in 2026.

47% said zero. 34% said one. 19% said two…

Well… nine days later, markets price a hike this Wednesday at roughly 90%.

What happened in between?

August PPI landed at +5.4% YoY on September 10. A full point above expectations.

$OIL spiked above $109 intraday the same day on Middle East supply fears.

Then August CPI arrived Friday: +0.4% for the month, 3.4% YoY. Core at 2.4%, actually the
lowest since March 2021. But headline is moving the wrong way, and this Fed has said prices come first.

The bond market repriced instantly. The 10 year yield near 4.94%. The 30 year at 5.37%, the
highest since 2007.

The S&P bounced +0.9% on Friday to 7,657. Retail sentiment on $SPY and $QQQ? Bearish, on high message volume.

The crowd was positioned for cuts, got a probable hike, and now distrusts its own bounce.

So what am I doing before Wednesday?

Nothing dramatic. And that is the point.
I don't trade Fed meetings. Rates move the discount rate. They change what a promise of 2030 earnings is worth today.

Order books live in a different world. Memory by $MU for instance is still sold out. Power queues are still years long.

The bottlenecks don't read the news… :)

A hike hits hardest where real cash flow is far away, where your mainly buying a narrativee. It hits least where profits are flowing today.

That is exactly why I own the suppliers who get paid now.

Risks, honestly: A surprise 50bp move would reprice everything, including my names. If oil stays above $100, the inflation story runs longer. High multiple cyclicals can fall 20% on a discount rate change alone. So expect volatility.

For years almost every Fed meeting arrived with near total consensus. This one arrives with a real split. Volatility around the decision is the normal outcome, in both directions.

Position so that Wednesday cannot force your hand. Get cash ready for buying discounted stocks.

Let's see what the week brings.

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

Gold miners just printed their best August since at least 1994.

The miners index rose 33% in one month. $AEM gained 40%. $NEM 35%. The metal itself added 10%.

And suddenly my feed is full of people selling AI winners to buy gold.

A few numbers before you join them…

$GOLD sits around $4,300-$4,400, up only a few points below 0% for the year. It still trades about 20% below its January record of $5,419.

So the metal went basically nowhere in 2026, the miners just had a historic month, and the crowd arrives now. Buying a hedge after its best month in three decades mostly means buying last month's chart.

On Friday last week gold fell 2% within hours of the hot jobs report. Hike odds up, gold down. Your crisis protection traded like a rate sensitive risk asset. Worth knowing before you hide in it.

A small ballast allocation you hold through everything? Reasonable. Panic rotation? That usually top ticks both sides.

I own a substantial amount as my "cash" position and emergency fund…It's my piggy-bank and once the market throws a sale I have enough funds to buy discounted stocks I believe in.

Do you hold gold next to your stocks? Or do you prefer to hold $BTC instead? Let me know in the comments.

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Evan @StockMKTNewz]

Nvidia $NVDA is reportedly considering investing up to $10 Billion into the Anthropic IPO

[Max @mrmagoon]

While everyone debated the memory top, China quietly took a tenth of the DRAM market…

The fresh Counterpoint numbers for Q2 landed this week:

CXMT reached 10% of global DRAM revenue. A year ago: 4%. YMTC hit 14% of NAND
revenue, moving past SanDisk's 11%. The three DRAM incumbents combined fell from 94% to
87% share.

The stocks noticed. $MU and $SNDK lagged last weeks Wednesday's tech rally on the news.

And the fear behind every thread I read: solar 2.0.

You know the story. China entered solar, scaled brutally, and prices never recovered. Western
manufacturers died. Now apply that template to memory and the whole supercycle thesis
wobbles.

I take the fear seriously. And then I look at what prices actually did while China gained that
share.

Blended NAND prices rose 55% quarter over quarter in Q2. $WDC reported its price per
terabyte up high teens percent year over year.

By Friday $SNDK closed at $1,740, up about
555% this year, the best performer in the entire S&P 500. $MU finished at $1,016.

Share shifted AND pricing accelerated. Both things are true at once.

Here is the layer that explains it: memory is not one market.

Chinese suppliers are scaling in commodity DRAM and consumer NAND, the price sensitive
end where cost wins. AI data centers by $NBIS or $IREN for instance run on the other end: HBM and enterprise storage, where qualification cycles take years, three companies produce at scale, and buyers care about supply security far more than price. $NVDA did not sign $279B of purchase commitments for the cheap stuff.

Solar had no such protected tier. Every panel was a commodity. That difference carries the
whole thesis.

What I actually watch from here:

CXMT's share of HBM, today near zero. If that number moves, the moat is draining. Contract
prices for 2027 capacity. And whether Chinese memory reaches Western flagship devices at
scale, which would free up incumbent supply.
The honest version: every commodity market China entered got cheaper eventually. The
question is the timeline, and whether the AI tier stays defensible long enough to matter.

10% is a warning shot, and it hit the exact week the sector made new highs. Cycles end when
nobody watches the back door. Consider it watched.

Which ones am I missing?

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

AI revenue up 221%… and this stock down 6%

Make it make sense.

$AVGO reported last Wednesday. The numbers were, objectively, monstrous.

Revenue $29.6B, up 86% YoY. AI revenue $16.7B, up 221% YoY. Up 54% in a single quarter.

EPS $3.32 versus $3.22 expected.

And the guidance was the sin: roughly $34.8B for Q4, while the street wanted about $35.0B.
A $250M shortfall. On a $35B quarter. That is 0.7%.

The stock fell 6% the next day to $357, sitting roughly 27% below its June high. Year to date it is up barely 1% while the semiconductor index gained over 60%.

Last week $MRVL raised its outlook by $1.5B and dropped 10%. This week Broadcom. The
invisible expectations bar keeps claiming victims.

Management laid out its AI revenue roadmap: about $58B this fiscal year. Around $115B in
2027. Around $230B by 2028.

Read that again. A company guiding to quadruple its AI revenue in two years just got sold for missing one quarter by 0.7%.

Morningstar called the decline an overreaction to conservative guidance and holds a $650 fair
value estimate. Retail agrees loudly: sentiment on the stock flipped extremely bullish with
message volume up over 1,000%.

But I want to be honest about the bear case too.

A roadmap is a promise, and $230B assumes hyperscalers like $GOOGL , $META and $AMZN keep writing checks at this pace
through 2028. Custom AI chips concentrate revenue in a handful of customers who renegotiate hard. And when a stock underperforms its own sector this much, someone big has been selling
into every rally.

My takeaway is the same lesson this market keeps teaching. The results set the floor. Expectations set the bar. And after two years of AI euphoria, the bar moves faster than the numbers.

For bottleneck investors the more interesting datapoint sits one layer deeper: those custom
chips Broadcom builds all need HBM from $MU or $SKHY advanced packaging and optical interconnects ( $CRDO ). The capex keeps flowing through the same chokepoints, whoever wins the socket.

I spend an unhealthy amount of time reading earnings calls. This one was a good quarter
dressed up as a bad day.

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[Max @mrmagoon]

162,000 new jobs

Consensus expected around 55,000

And the market's reaction? Red.

Welcome to the strangest logic in investing: good news for the economy became bad news for stocks.

Friday's jobs report tripled expectations. Unemployment held at 4.1%. Even July got revised from minus 23,000 to plus 21,000. The labor market cracks everyone feared a month ago? Largely revised away.

Sounds great. So why did the S&P close down 0.4% at 7,718?

Because of what strong data means for the price of money.

Odds of a September rate hike spiked toward 65% right after the print, then settled near a coin flip by Friday's close. The 2 year Treasury yield hit 4.374%, its highest level since January 2025. The 10 year touched 4.78%.

A hot economy gives the Fed under Kevin Warsh room to fight 3.4% inflation with a hike. A hike raises the discount rate on every stock. High multiple AI portfolios feel that first and hardest.

And here is the part I find fascinating: retail refuses to believe it.

A StockTwits poll this week found 47% expect ZERO hikes in 2026. Another 34% expect one. Only 19% expect two. The crowd sits positioned dovish while the bond market prices a coin flip. Someone will be wrong. Soon.

The calendar that decides it:

PPI on Thursday. CPI on Friday. Fed decision September 16.

So what am I doing with my AI heavy portfolio into that gauntlet?

Mostly what I always do…

Rates move the multiple on my stocks. The order books at the bottleneck companies run on physical shortages, and those shortages do not care about 25 basis points.

But I respect the other side: if a real hike cycle starts, valuations compress first and ask questions later. My protection stays boring. Position sizing. A cash buffer. No leverage.

One thought I keep coming back to: the Fed debating a HIKE is itself evidence the economy can carry this AI buildout. The scary scenario was never strong jobs. It was the cracks.

What's your read for September 16: hike or hold? Tell me below.

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

$GOOGL $META $NVDA $BTC $GOLD

[Max @mrmagoon]

Everyone watched Nvidia's revenue last week

I went down a rabbit hole in the footnotes instead…

Buried in the filing sits one number that tells you more about the next two years of this cycle than the entire income statement.

Purchase commitments -> Last quarter: $119B, this quarter: $279B.

Nvidia's contractual promises to its own suppliers more than doubled in three months. The bulk of it, per the company, secures one thing: memory. Management said it directly on the call: memory supply is a major constraint, expected to persist through at least fiscal 2028.

Now here is the part that makes this week fascinating…

You would expect memory stocks to explode on that news! The opposite happened.

Last Monday brought a report that $AAPL may source some memory from Chinese suppliers. $MU fell 7%. $SNDK fell 9%. $WDC fell 7%. All in one single day last week.

Thursday, right after Nvidia's print confirmed the shortage: $MU down another 3% to $913.62. $WDC down 4% to $449.18. $SNDK down 1% to $1,479.23.

And retail sentiment on $MU? A bit bearish.

The crowd is going quiet on memory in the exact week the biggest customer on earth signed a $279B receipt.

So which alpha do I trust? Orders and contracts!

A purchase commitment is a contract. It sits in Nvidia's filings as an obligation. Nvidia is reserving memory capacity years ahead because it cannot ship its next generations of systems without HBM. That is real demand with a signature on it.

A red week in the stocks is positioning. After runs like +229% YTD for $MU, +532% for $SNDK and +172% for $WDC, funds take profits, momentum traders rotate, and every headline becomes a reason to sell first and read later.

Prices move on positioning within weeks. Businesses move on contracts over years. Confusing those two timeframes is the most expensive mistake in cyclical investing.

The Apple story deserves respect, to be clear. If Chinese memory ever reaches flagship phones at scale, that adds supply to the consumer segment. But HBM for AI data centers is a different pool. Three companies produce it at scale. The buyers are price insensitive. And the new fabs answering this shortage only deliver meaningful supply from around 2029.

One more datapoint from the same stretch that got almost no attention: Micron announced Micron Research Labs, a $10B research hub in Boise built over the next decade, described as the first dedicated memory research center of its kind in the US. Groundbreaking in 2027. Jensen Huang and Tim Cook publicly endorsed it.

Step back & look at the sequence…

The customer more than doubles his supply commitments. The supplier invests $10B in long horizon research. The CFO of the most valuable company on earth calls memory the constraint. And the crowd goes bearish on low volume.

Memory names on my radar: $MU $SNDK $WDC $SKHY

The honest risk side: Memory might be still cyclical. Every supercycle in history eventually ended in oversupply. Contract prices can decelerate long before the headlines turn. Purchase commitments can be renegotiated in a genuine downturn. And after these YTD runs, a 30% drawdown is statistically normal. It will feel terrible anyway.

Zoom out… Capital keeps flowing toward the next constraint. Compute was first. Memory carries the cycle now. The $279B tells you Nvidia agrees with that map.

I am not changing anything based on one red week. The thesis gets reviewed when the contracts change, the supply timeline changes, or the customers stop complaining about prices.

None of that happened last week

What are your thoughts on memory stocks?

Save this post and revisit it later!

—

I’m running a small private Telegram group of high-quality eToro investors.

If you want to join, visit my X account (link on eToro profile). I posted a link to the waitlist for the group there.

Let’s keep building growth,
Max

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

[chl$ @chelsssseeeea]

Maybe one of the biggest mistakes modern culture made was treating responsibility as the enemy of freedom.

Marriage restricts you. Children restrict you. Owning a home restricts you. Caring for family restricts you. A career restricts you. Community obligations restrict you.

Of course they do. Anything meaningful places demands on you.

But if you remove every obligation in pursuit of maximum freedom, eventually you’re left with a life where nobody needs you, nothing depends on you and there’s very little bigger than yourself.

We talk constantly about wanting meaning while systematically avoiding the exact commitments that tend to create it. And I guarantee you there’s going to be a lot of sad and lonely people in our aging generation because of this nearsighted pursuit.

[Anthropic @AnthropicAI]

Many drugs work by binding to a specific target in the body and blocking or changing what it does. An important first step in the drug development process is designing a molecule that can bind tightly to its target. Traditionally, that's meant weeks or months of expert work per target, sifting through a large number of candidates to identify the few that work.

We wanted to test if Claude could successfully design novel protein binders from scratch (also called de novo design). With a protein design prompt written by a human expert, Claude autonomously designed protein binders against 14 out of 15 targets.

We then worked with Adaptyv Bio and Twist Bioscience, who independently built and tested the proteins Claude designed.

[Max @mrmagoon]

Happy Sunday! I am preparing a new post on eToro all about my/hardest learnings in investing and would love to include your experiences (incl. your ticker if you want). Would someone like to collab? Then just share your answer to one or some of these questions and attach your etoro handle if you want to be tagged:

1. How did an emotional decision change the way you invest?

2. Which moment changed your definition of "risk"?

3. How do you determine your investment has failed?

4. What do you tell yourself when the market goes into a steep correction?

5. When was the last time you sold a winning investments and regretted it?

6. What’s an uncomfortable truth about your investments?

7. When was the last time a trade permanently changed the way you look at money?

Would love to hear your thoughts on these and feel free to add another question, if I am missing one.

[Max @mrmagoon]

The US economy lost 23,000 jobs in July…

The S&P 500 closed at a record high the same day. 7,757.64.

How does this make sense?

I saw versions of this question all last week: how can the market celebrate a shrinking labor market by pumping more money in the market?

Here is how I would explain it...

Markets price two things: future earnings and the interest rate used to discount them.

The jobs report hit both. Expectations were around plus 83k jobs. Instead we
saw -23k! May & June were revised down by another 103k combined. Wages grew 3.2% YOY, the slowest pace since May 2021.

Bad for earnings you would guess, right? Not so much.

What actually moved the market was speculations on upcoming interest rates. The Fed currently sits at 3.50% to 3.75% interest, and markets had priced a real chance of a hike in September. The market was a bit afraid.

But what went down is, that those odds fell drastically once it became clear that the inflation rate would meet expectations. That alone was a bullish signal.

Less probability of a rate hike is a basically a lower discount rate on stocks and therefore is… a bullish signal.

So a lower discount rate lifts the sentiment for every future cash flow, as money becomes cheaper. That’s especially good for growth stocks, which are more dependent on debt.

That is the whole trick. Friday was a rate story wearing a jobs costume.

Unemployment fell to 4.1%. Sounds good. But it fell because 264k people completely left the labor force. Participation dropped to 61.4%, the lowest in five decades outside the pandemic.

And "bad news is good news" has an expiry date. It works while weakness stays mild. Once job losses start eating consumer spending and corporate revenue, bad news becomes just bad news and the market will turn.

My read as an investor: I never position for a single print. But I do watch September 15 and 16, the next Fed meeting.

This is not investment advice, for education purposes only | Capital at risk | Past performance does not guarantee future results

$VOO $QQQ $GOLD $BTC $ETH