The US government is spending $7.6 trillion on AI infrastructure by 2031. Microsoft has $80 BILLION in unfulfilled Azure orders... NOT because of demand, but because they simply CANNOT get enough GPUs, power, or land. GPUs have a 36-52 week lead time. Blackwell is sold out through mid-2027. TSMC's 3nm wafers have demand exceeding capacity by 3x. The hyperscalers can't build fast enough. Enter the NEOCLOUDS. Pure-play AI cloud providers that do ONE thing, and one thing ONLY: rent GPUs. The neocloud market went from $0 to $25 billion revenue in just 3 years. It's projected to hit $400 billion by 2031. The biggest question is this: WILL THOSE PREDICTIONS ACTUALLY COME TRUE? If so, this industry is set to explode. If not, it may crumble. Three companies are leading the charge: $NBIS $IREN $CRWV Along with a slew of small and microcaps: $BRUN $WYFI $SHAZ and more... This is the biggest infrastructure buildout since the internet itself. The last time we saw spending at this scale, it created Amazon, Google, and Meta. This is the infrastructure layer of the AI revolution. Don't sleep on it. Timestamps: 0:00 Intro 0:40 The Black Swan 1:20 What is a Neocloud 3:40 Iren Overview 7:15 Nebius Overview 10:22 Corweave Overview 11:26 The Small Fish 12:47 The Black Swan That Wasn't 16:58 Final Thoughts

365 days ago, I decided to go all in on making content about stocks. 150K+ combined followers later... I thought I'd share my 3 best and 3 worst longs so far. And once you read my L's, you're going to realize why I chose this path. 👇Based on the date I dropped a bull thesis (YT, IG, substack mix) Winners: 1) $MU +431% 2) $AAOI +201% 2) $AMD +199% Losers: 1) $ORBS -88% 2) $FIG -76% 3) $AVEX -44% Notice how much more money could've been made on the best stocks versus lost on the worst stocks? That's why we play this game. James Anderson from Bailey Gifford said it best. “The asymmetric payoff structure, you can make far more if you're right about a stock than you can lose if you're wrong, is the fundamental attraction of investing in equity markets". It's been a fun ride so far, and we never know where markets will go next. But massive shoutout to my partner @KawzInvests and new teammates @LanceTeren @calioncap @yvesstocks. Plus the great traders/friends I now talk to almost everyday @BULLOFBRITAIN @CKCapitalxx @OptimusDelta @PhotonBull @charaninvests

The Anthropic news is SOO bullish for Chinese AI btw One government letter forced Anthropic to pull its two most powerful models in 90 minutes. Closed-source US models are a single point of failure. Chinese labs ship open-source models you can self-host on your own infra. HKEX: 2513 — Knowledge Atlas (Zhipu) -> China's leading pure play AI lab out of Tsinghua, IPO'd jan 2026 -> US cut off their nvidia chip access entirely in jan 2025. -> instead they rebuilt on chinese hardware and shipped GLM-5, a 744B parameter model now beating claude on coding benchmarks. sanctions made them more efficient than labs -> cloud ARR hit $250M in march, targeting $1B by year end. stock up 1,600% since IPO HKEX: 00100 — MiniMax -> shanghai lab IPO'd jan 2026, oversubscribed 1,837x -> M2.5 is beating claude opus on SWE-bench verified, the coding benchmark that actually matters -> Hailuo AI their video gen product is going viral globally, 212M users across 200+ countries -> ARR went from $150M in feb to $300M by end of may. targeting $1B EOY $BABA — Alibaba -> Qwen 3 is kinda competing with the best US frontier models and most people haven't priced this in -> IF(very big if) enterprises rotate out of US AI they need somewhere to land. Alibaba has the models and cloud infra to absorb it at scale

$CBRS IPO this week and I think it goes nuclear Cerebras built what nobody thought was possible. One chip. An entire silicon wafer. No stitching together hundreds of GPUs, no bottlenecks, no bs. Up to 15x faster inference than Nvidia’s best. Groq has been getting all the hype for fast inference. Cerebras makes Groq look slow. OpenAI just signed a $20B deal with them -> $510M in revenue last year -> every major tech company on the planet is trying to find an alternative to Nvidia and this is the only thing that actually delivers. “Cerebras is the first architecture in decades that actually challenges the GPU at its own game” -> Dylan Patel, SemiAnalysis They are IPOing at the perfect time. Just look at $SMH. AI infra spend is going vertical. The window is open right now. IPO range $115-125. I expect it opens above $170. Already trading above $250 on Hyperliquid. Fills will probably be shit given how oversubscribed (~20x) the IPO is - I am watching $WYFI and $VICR as sympathy plays

Worst price action in months. Best news day the space sector has ever had. $ASTS: -30% from ATH $RKLB: -30% from ATH $RDW: -35% from ATH $PL: -15% from ATH Today SpaceX confidentially filed for the largest IPO in history. $75B raise. $1.75T valuation. Polymarket has June at 60%. Every fund manager who has ignored space now has to explain why. That flow has to go somewhere. $SATS Looks interesting. EchoStar holds a ~2.8% stake in SpaceX, received as equity consideration in its spectrum asset sale. NAV is estimated at $150-180/share 30-50% above where it currently trades with the IPO as the direct catalyst to close that gap. The sector is being sold at exactly the wrong time to be short it.

This is bullish af

Anthropic is a ~$350B company. but with one tweet they can: erase ~$400B from SaaS erase ~$50B from cybersecurity wipe ~$30B off IBM in under 6 hours all they need to do is: step 1: (hypothetically) have an employee load puts on “legacy incumbents” step 2: announce a new Claude capability step 3: watch the whole sector tank just one tweet and trillions start questioning their moat actually insane

NAND giants like Samsung, SK Hynix, $MU, and $WDC via the Kioxia JV got crushed in the 2022 to 2023 downturn. Prices collapsed nearly 70 percent and the industry was forced into roughly 40 percent capex cuts. That pain permanently changed behavior. Now AI is driving more than 50 percent annual NAND demand growth in data centers. The industry has shifted from chasing volume to enforcing price discipline. There is no incentive to overbuild and destroy margins again. Supply stays tight because new fabs take years to matter. Projects like Micron Boise and SK Yongin will not meaningfully impact capacity until 2028 at the earliest. Meanwhile hyperscalers are locked into long term agreements through 2029 securing allocation at premium pricing with projected margins in the 65 to 70 percent range. This is not greed. It is strategy. Prioritizing profitability over volume creates sustained undersupply and turns NAND into a margin fortress for the next several years. The numbers already confirm it. $WDC NAND revenue up 94 percent year over year with margins at 52 percent and rising. $MU showing over 60 percent growth while also owning the HBM angle for AI. AI makes this a structural multi year shift. Long $MU $SNDK

BULLISH $MU $SNDK

$LTRX - The Hidden Anduril/ONDAS Play Trading at a Fraction of Comps The Anduril supply chain keeps expanding. Here's one nobody's talking about. Lantronix just dropped a partnership with Safe Pro Group this morning. Most will scroll past it. Here's why that's a mistake. The Connection Nobody's Made: Safe Pro Group is the company Ondas Holdings poured $14M into back in October. $ONDAS now has a $5B market cap trading at 100x+ sales. But here's the real alpha: In December, Trillium Engineering selected LTRX Edge AI modules to power their gimbaled imaging systems for military drones. Pull up the Anduril Ghost spec sheet. The sensor listed? Trillium HD45. Connect those dots. LTRX appears to be in Anduril's supply chain. And now they have a direct relationship inside the ONDAS ecosystem too. The Valuation Gap Is Absurd: $ONDS - $5.0B market cap, ~$50M rev, 100x P/S $RCAT - $1.6B market cap, ~$40M rev, 40x P/S $KRKNF - $1.6B market cap, ~$100M rev, 16x P/S $OSS - $250M market cap, ~$30M rev, 8x P/S $LTRX - $270M market cap, $118M rev, 2.3x P/S $LTRX has more revenue than $ONDS, $RCAT, and $OSS combined. Trades at a fraction of the multiple. Why the Disconnect? $KRKNF is known as the Anduril sonar/battery supplier. That connection drove a 250%+ run. $LTRX has the Anduril imaging connection through Trillium and nobody is talking about it. They also have Red Cat for the Army SRR program. Three nodes in the defense drone ecosystem. One stock trading at legacy industrial multiples. The Setup: 5x P/S (half of $KRKNF) = $590M = 2x 10x P/S (still below comps) = $1.2B = 4x

Nancy Pelosi just disclosed roughly $69 million in stock trades. She sold $50 million in Apple while buying call options in Google, Amazon, Apple, and Nvidia. What's striking isn't just the amount, it's that her trading performance is statistically impossible to explain through normal investing skill. Over the past decade, her portfolio returned 838% versus the S&P 500's 256%. She's won 86% of her trades, accumulated $160 million in realized gains, and built a $572 million portfolio. These numbers suggest consistent access to information the public doesn't have yet. This is exactly what members of Congress possess through early knowledge of policy changes and regulatory decisions. Trading on nonpublic information is supposed to be illegal, which is why Congress passed the STOCK Act in 2012. The problem is the law is completely toothless. Violations come with a $200 fine for first time offenders, and not a single member of Congress has ever been prosecuted in 14 years. When the Treasury Department investigated which members had actually paid their fines, they found "no matches." Enforcement isn't even being tracked. In 2020 and 2021 alone, 57 members and 182 staffers were late filing required disclosures with zero consequences. The core issue is an unavoidable conflict of interest. Members vote on tax policy, healthcare regulation, antitrust enforcement and tech sector rules. When they hold millions in stocks tied to companies affected by these votes, they face direct personal incentive to steer legislation toward enriching themselves rather than serving constituents. It's impossible to be objective when your net worth depends on how you vote. The encouraging part is that solutions exist with massive public support. Eighty-six percent of Americans support banning congressional stock trading. Representative Bryan Steil introduced the Stop Insider Trading Act, which would ban new stock purchases by members, require advance notice of sales and increase penalties to 10% of transaction value. Senators Gillibrand and Hawley proposed an even stronger version requiring complete divestment into index funds and blind trusts. Pelosi can't be stopped individually because she's not technically breaking current laws. What needs to happen is Congress has to change the rules themselves, ban stock ownership for member. Also, create real penalties, allow real time disclosure instead of the current 30 to 45-day delay, and rebuild the public trade database. Until that happens, members will keep enriching themselves while making decisions that should be based entirely on what's best for their constituents.

Denmark's doing something worth paying attention to. Their US Treasury holdings just hit a 14 year low at $9 billion, down 30% over the past year. But the real story is Akademiker Pension, a major Danish pension fund managing $25 billion in assets, which is dumping all $100 million of its US Treasury position by end of January.​ Why? The fund's basically saying US government credit is broken. Their chief investment officer told Bloomberg straight up: "the US is basically not a good credit and long-term the US government finances are not sustainable." They're citing concerns about fiscal discipline, weak government finances, and rising credit risks under Trump's policies. The Greenland tensions didn't help either, he admitted that "certainly did not simplify our decision-making process." This matters because it's not just one fund. Other Danish pension funds are also reducing or exiting their Treasury holdings. It signals something bigger, institutional investors are reassessing whether US government debt is still the safe haven investment everyone's assumed for decades.​ The US needs about $1 trillion annually from foreign investors to fund its deficits. If demand keeps cooling, Treasury yields have to rise to attract buyers. And when Treasury yields rise, everything else gets more expensive mortgages, car loans, credit cards, business borrowing. It's inflationary and hits regular Americans in the wallet.​ Treasury Secretary Bessent dismissed it as irrelevant and technically he's right, $100 million is nothing in a $30.8 trillion market. But that's missing the point. The question isn't whether Denmark breaks the market today. It's whether this is the start of a confidence erosion among foreign central banks and institutions. If Japan, Switzerland, and other major holders start following Denmark's lead, that cumulative effect actually matters for US borrowing costs.