The Edge Street

Most people think investing is about picking the right stocks. It is actually about avoiding the wrong decisions. Not panic selling in March 2020 when the market fell 34% in a month. Not sitting in cash in 2023 waiting for the recession that never came. Not selling during the Iran war fears last Monday when the Dow fell 557 points. The $SPX is up 8% since January. Up 40% in 12 months. Six straight weeks of gains. Every single one of those gains was available to anyone who simply did nothing. The best investment strategy most people will ever find is boring, consistent, and requires almost no skill. Buy quality. Hold through the noise. Let time do the work.

2 year payback on $50B of capex is why hyperscalers keep raising their spending guidance every single quarter. When your return on investment is that clear and that fast, the only rational move is to build more. The constraint isn't demand or economics, it's power and physical capacity. That's why IREN, CEG and the power infrastructure plays exist. The bottleneck moved from chips to electricity.

Back-of-envelope numbers for 1 gigawatt data center: All-in Capex: ~$50 bn Enterprise revenue generated: ~$25-30 bn/year Electricity cost: $1-2 bn/year ~2 year payback. The boom is real.

@cmsinvests And the infrastructure buildout hasn't even hit full speed yet. $750B in capex this year alone across just four companies. Most people are watching the application layer. The real money is in the pipes, the power and the memory that everything runs on.

“Why is the stock market rallying so hard?” Because Wall Street finally realizes AI isn’t just some ChatGPT bot writing essays. Most people still don’t understand what’s happening underneath the surface. AI requires an entirely new global infrastructure buildout: • GPUs → $NVDA • Memory/HBM → $MU $SNDK • Servers → $DELL $SMCI • Networking → $ANET • Power demand → $VRT • Cloud hyperscalers → $MSFT $AMZN $GOOGL Trillions of dollars are about to be spent rebuilding the backbone of the internet for AI. And the crazy part? The average person still thinks AI peaked with Grok images…

@calvinfroedge About 89 days of 1% gains to get from 41% to 100% if everything else stays flat. But the more interesting question is what happens at 60%, 70%, 80%. At some point passive index funds mechanically have to keep buying the winners because they're the index.

Here's a brain teaser for you If tech / the AI trade is 41% of global market capitalization today How many days of going up 1% until it's effectively the entire market capitalization? Assume continued concentration of gains (i.e. rest of market flat)

The credit provisions point is the most underappreciated part of the whole thesis. $1.2B in provisions looks scary on a headline basis. But if those loans are pulling millions of first time credit users into the ecosystem and the cohort data shows predictable payback, you're essentially capitalising customer acquisition through the P&L. Amazon did the same thing with Prime. The subscription looked like a cost centre. It turned out to be the stickiest retention mechanism in retail history. MELI is building that same flywheel across a region that's still in the first innings.

One thing I think is fascinating about $MELI is that the financials are becoming harder to understand exactly because the business is becoming stronger strategically. That’s quite the conundrum… A simple marketplace is easy to analyze. You look at GMV growth, margins, advertising revenue, etc. But once you add lending, credit cards, fulfillment, logistics, acquiring, and 1P inventory, the numbers naturally become complex because the business itself becomes more powerful and vertically integrated. I also think investors underestimate how much accounting can hide what is actually happening underneath. Free shipping looks terrible for margins today, but if it permanently changes customer behavior and increases purchase frequency for years, the long term economics can end up being incredible. The same thing applies to credit. Provisions rising from $603m to $1.2b looks scary up 100%, and it should be monitored carefully. But if those loans help pull millions of users deeper into the ecosystem while underwriting remains reasonably disciplined, then part of those “losses” are really customer acquisition costs disguised as credit expenses. Another thing worth thinking about is how different $MELI opportunity is compared to a US ecommerce company. In the US, ecommerce and digital payments are already mature. In large parts of Latin America, many people are still early in adopting online shopping, digital wallets, credit, and banking. That means $MELI is not just fighting for market share. In many cases they are helping create entirely new consumer habits. Those kinds of markets can become much larger than people initially expect because the pie itself keeps expanding. I also think the market sometimes overreacts to margin compression while underreacting to ecosystem strengthening. Warehouses, fulfillment, fintech, acquiring, ads, and credit all reinforce each other. The more products a customer uses, the harder it becomes to leave the platform. That is why $MELI is becoming harder to value. It is no longer just an ecommerce business. It is slowly becoming financial and commerce infrastructure for an entire region. 🌹

Hard to argue with that pick. Three completely different businesses all compounding at the same time. AWS, advertising and retail each big enough to be standalone trillion dollar companies. $GOOGL is also a top pick If i had to choose an alternative. DeepMind, Waymo, TPUs, cloud and the best search moat ever built.

If you could only invest in 1 company for the next 20 years, which would it be? I’ll go first: $AMZN

Growing faster than the index, cheaper than the index, and down on the year while everything else rips. The only reason it's not getting credit is the capex fear. $125-145B in spending and the market is still waiting to see the return. But the return is already showing up in the revenue numbers. 33% growth on a $56B revenue base doesn't happen by accident. Yeah it's becoming obvious.

$META is down over 6% this year, while the S&P 500 is up 8%. Meta Forward P/E: 18.35 S&P 500 Forward P/E: 21.4 Meta projected 3 YR EPS CAGR: 20.5% S&P 500 projected 3 YR EPS CAGR: 15% Is $META becoming obvious at this point?

MU at 0.4x PEG while supplying memory for the biggest compute buildout in history is genuinely hard to explain. NVDA and AVGO both under 1x PEG tells you the growth is still outpacing the valuation expansion. That's not what expensive looks like. INTC at 2.8x after a 500% run is the one that actually deserves scrutiny now.

SEMICONDUCTOR STOCKS BY PEG RATIO PEG < 1 usually means mispriced growth PEG > 2 starts to push into the danger zone Here’s how they stack up: • $INTC ~2.8x • $LRCX ~2.0x • $KLAC ~2.0x • $AMAT ~2.0x • $ASML ~1.7x • $ALAB ~1.6x • $ARM ~1.5x • $ANET ~1.5x • $LITE ~1.3x • $TSM ~1.1x • $CRDO ~1.0x • $NVDA ~1.0x • $COHR ~0.9x • $AVGO ~0.9x • $AMD ~0.7x • $SNDK ~0.7x • $MRVL ~0.7x • $AAOI ~0.6x • $ON ~0.5x • $MU ~0.4x

🚨 WARREN BUFFETT STEPPED DOWN AS BERKSHIRE CEO AFTER 60 YEARS. He didn't time the market. He didn't trade options. He didn't chase the hottest sector. He bought great businesses at fair prices. He held them through crashes, corrections, wars, recessions, and pandemics. And he let time do the work. $10,000 invested in Berkshire when he took over in 1965 — worth over $350 million today. The most powerful investment strategy in history isn't complicated. It's just very hard to be patient enough to execute it.

$136B into money markets the same week stocks hit all time highs. That's not retail panic selling. That's institutions taking chips off the table after a 17% rally in 30 days and parking cash somewhere it earns 5%. The market can keep going up while this happens. But when that much capital is sitting on the sidelines, it tells you conviction in the rally isn't unanimous.

BREAKING: Money market funds posted +$136 billion in inflows last week, the largest weekly intake since January 2026. This is also the 2nd-largest weekly inflow since the beginning of 2025. This follows -$175 billion in outflows in the preceding week, the largest weekly withdrawal on record. As a result, the 4-week average of outflows stands at -$45 billion, the 2nd-largest on record. Meanwhile, investors allocated +$25.9 billion to bonds last week, with Investment-Grade bonds alone attracting +$16.4 billion, the largest weekly intake since January 2026. Investors are shifting capital after a historic run.

CPI on Tuesday is the one that matters most right now. Sticky inflation kills the rate cut narrative completely and the market is already pricing in barely any cuts this year. A hot print and yields spike, growth stocks get hit. Big week. Everyone's been ignoring macro while semis rip. Tuesday might force them to pay attention again.

Key Events This Week: 1. April Existing Home Sales data - Monday 2. April CPI Inflation data - Tuesday 3. April PPI Inflation data - Wednesday 4. OPEC Monthly Report - Wednesday 5. April Retail Sales data - Thursday 6. April Industrial Production data - Friday More crucial inflation data is ahead of us.

Analyst price targets are worth something but they're not the whole story. The real question is whether the AI upsell actually moves revenue per seat. They've been talking about NowAssist for two quarters and the market wants proof, not promise. At $91 the risk reward is interesting. But the next earnings print needs to show the attach rates are real or that 50% upside stays theoretical.

ServiceNow is $91/share. Average analyst PT: $142/share. That suggests Wall Street thinks there is over 50% upside in this stock. Are you buying? $NOW

The energy layer is the most underappreciated part of the entire AI buildout. Everyone's focused on chips and data centers. Nobody's asking where the power comes from to run them. CEG with the Three Mile Island restart and a 20 year Microsoft contract is the cleanest play on this list. Nuclear baseload that runs 24/7 regardless of weather is exactly what hyperscalers need and can't get from solar or wind.

10 WAYS TO BUILD AN AI POWER PORTFOLIO 1. $OKLO effectively building the “local nuclear plant” the AI economy will require by placing reactors directly next to data center campuses for 24/7 onsite generation. 2. $BE fuel-cell onsite power play helping data centers bypass the grid with dedicated energy for AI clusters with product backlog up 250% YoY to $6B. 3. $CEG nuclear baseload backbone of the AI era with a 20-year $MSFT PPA tied to the Three Mile Island restart to supply the 24/7 carbon-free power. 4. $VST hybrid power engine of AI combining nuclear, gas & storage with a 20-year $META agreement covering 2,600+ MW across three nuclear plants. 5. $GEV industrial supplier rebuilding the U.S. grid providing the turbines, transformers & hardware every AI-driven upgrade cycle depends on with $163B in backlog. 6. $VRT infrastructure gatekeeper for AI compute controlling the cooling & power systems that $NVDA class clusters cannot run without with Q1 backlog up 80% YoY to ~$12.5B. 7. $EOSE long-duration storage solution for a grid under strain helping utilities smooth volatility as AI demand overtakes supply. 8. $NEE clean-energy arm of the AI buildout with largest renewable development pipeline in the country positioned directly into data center load growth. 9. $LEU only U.S. source of HALEU fuel making it essential for powering the modular reactors needed around future AI campuses backed by ~3B DOE contract. 10. $UUUU secures the domestic uranium supply chain by turning nuclear fuel into a national-security asset for the AI age.

Jensen isn't just building the best AI chip company. He's building stakes in every company that depends on his chips succeeding. If the AI buildout continues, Nvidia wins on the GPU revenue. If it accelerates beyond expectations, they also win on every equity position in that list. He's got an ecosystem.

$NVDA has committed more than $40B to equity investments this year alone: • OpenAI $30B • $GLW ~$3.2B • $IREN ~$2.1B • $MRVL ~$2B • $LITE ~$2B • $COHR ~$2B • $CRWV ~$2B • $NBIS ~$2B Nvidia is positioned across every layer of the AI buildout from silicon & networking to optics, power, neoclouds & the model companies running on top. The whole stack runs through Jensen.

🚨 THE MARKET IS PRICING $AMZN AMAZON AS ONE COMPANY. IT'S ACTUALLY SEVEN. AWS alone — $110B revenue, 38% operating margin, growing 17% YoY. Valued at a 30x multiple, AWS by 2036 would be worth over $7 trillion. Amazon's current market cap: $2.1 trillion. So you're buying AWS at a significant discount. - And getting the world's largest e-commerce platform for free. - And the world's largest digital ad business. - And Prime — 200 million subscribers. - And Alexa. And Kuiper. And a healthcare division. Every single one would be worth billions standalone. The market sees one company. The opportunity is in seeing seven.

Insider selling at ATHs isn't automatically bearish though. Executives have compensation packages loaded with stock. They sell for diversification, tax planning and liquidity reasons regardless of what they think about valuation. The metric worth watching is the buy to sell ratio. If insiders are buying alongside the selling, the signal is mixed. If nobody is buying at all, that's the real warning sign.

WARNING 🚨 Insiders continue dumping stocks at a rapid pace while retail investors keep buying at all time highs. This is how bubbles form.

The concept is right but 10% is too high for most people starting out. 5% in high conviction bets, the rest compounding quietly. The gamble fund only works if losing it all doesn't derail the main portfolio. The real discipline is not raiding the boring portfolio when the gamble fund hits.

the underrated financial advice is to build a gamble fund with 10% of your annual income the motive of gamble fund should be investing in 10x bets including hot stocks, meme coins or small business you will make 100% of your annual income in some years by risking small amount

The "software is dead" narrative and a $1T target from the same company at the same time is the exact tension worth sitting with. AI either kills the workflow layer or makes it 100x more valuable. Jensen Huang said the latter publicly. The 14% selloff after a beat says the market isn't sure yet. McDermott has hit every target he's ever set at ServiceNow. That's the only reason the $1T number deserves to be taken seriously.

Wait WHAT?!?! ServiceNow $NOW is guiding to become a $1 TRILLION dollar company eventually while the market cap ain’t even $100B right now. If they get even close to that guide within the next decade returns are exceptional. Software is dead though according to the market. I am unsure about software myself. Idk boys.

🚨 MICHAEL BURRY JUST COMPARED THE AI RALLY TO THE FINAL STAGES OF THE DOT COM BUBBLE. Here is what he got right. And what he is missing. He is right that sentiment is extreme. The Philadelphia Semiconductor Index is up 65% in 2026 alone. Retail is piling in. Valuations on some names are pricing in perfection. That is all true. Here is what is different from 1999. In 2000, the companies leading the bubble had almost no revenue. Pets. com. Webvan. Kozmo. The entire narrative was built on future potential with nothing underneath it. Today $NVDA Nvidia generated $216 billion in revenue last year. $GOOGL Alphabet grew cloud revenue 28%. $META Meta posted a 41% operating margin. $AMD crossed $700 billion market cap on real earnings growth. The revenues are real. The earnings are real. The AI capex is real and accelerating. Burry may eventually be right on timing. He was right on housing in 2005 and spent two painful years being called an idiot before the crash came. But calling this a repeat of 1999 ignores the single most important difference. This time the companies are actually making money.

When Jensen publicly endorses a company in this environment it's not just a compliment, it's a signal about where Nvidia is deploying infrastructure spend. IREN at $60 with the NVDA partnership already signed is the most concrete of the three. The revenue is contracted, the power is coming online, the relationship is real. Worth paying attention to.

The $NVDA CEO has been literally telling you what to buy… In 2025 Jensen called out: $NBIS at $21 & is now up 840% $APLD at $3 & is now up 1,400% $TSM at $180 & is now up 135% $MU at $86 & is now up 770% Jensen is now calling for these 3 companies to squeeze in 2026: $NOW at $90 $CRWV at $114 $IREN at $60 These names are most likely the next 10x setups. Don’t miss out…