"Are we in an AI bubble?" send them this 👇 May 2020. Nvidia ships a chip called the A100. For two years now the bears have called these things melting ice cubes. Useful for three years, then landfill. Cassandra built an entire short thesis on exactly this. His argument: big tech depreciates GPUs over five or six years when the real life is two or three, and that little trick hides **$176 billion** of expenses between now and 2028. It's a good story. Clean, simple, easy to repeat... Then last week CoreWeave got on their earnings call...

306 🐊🐊🐊 Who remembers this chart? 59 releases to ATH

Closing up September with a modest 5% gain, taking our portfolio to new ATHs and a nice 18% YTD with a 4 out of 10 risk score (read = index fund like). If treasury yields stop the bleeding and do a U turn (somehow, maybe with a buyback bazooka by the House), then: - more liquidity in circulation = hard assets appreciate - lower yield = higher DCF and therefore higher valuations for equities - investors make a buck on their treasuries = go shop around for other assets such as equities - equity investors make a buck on their appreciated equities = go shop around for crypto - and so on down the risk curve with watches, collectibles etc. Wish you all a fantastic Q4. Watch Japan, yields and self fulfilling prophecies.

This and the prophecy made me buy some 100k strike end 2026 expiry BTC calls. Sorry everyone

- 6 ⌛️

$SUI Basecamp = guaranteed pump

Handed over the keys to my agent, follow at your peril. Interesting experiment though, they were pretty early in the AI energy trade.

~500% cumulative returns and ~30% annualised since people started copying me on eToro in 2020. All documented and transparent. 2019 was another great year, but I wasn’t part of the programme back then, so it wouldn’t be fair to include it. That’s what I like about eToro: the performance is public. I can’t just post the winners. All my trades are there for everyone to see (and judge :p). Unfortunately I don’t have a finance degree (I do have a master’s in business, though), so I couldn’t even get considered for a job in finance. eToro gave me a platform to show what I could do, tell my investment thesis over the years and I couldn't be more grateful for it. My portfolio includes a mix of indices, AI, energy, emerging markets and select crypto, with the allocation changing over the years. Long only. NFA. Not a solicitation to invest.

2025 btw

Top blasting $ZEC ✅

Idk why but sometimes I find myself thinking about this legend. “Show me the incentive and I will show you the outcome”. Truly unmatched.

305🐊 60 more releases to go for the new ATHs

I normally steer away from shorting anything. If I need to hedge I buy puts. But there's this project that is being propped up by market makers and that should be trading at 1/10th of the current mcap So I short it and farm the funding rate from the market makers. And I am up 186% on the short itself. Sometimes it's just about knowing enough about new projects to leverage market inefficiencies and market making behaviors.

304 ⏰

it's always darkest before the dawn

303 😅

Was too short of a tweet so let me elaborate... Markets either wear you out or scare you out. $CRWV $NBIS and the likes drop double digit overnight, but the renting prices of "old" chips keep rising... 1)what? H100s went from $1.70/GPU-hour in October 2025 to $2.35 by March 2026 (semi analysis report) and now over $3 (from Ornn data). Neoclouds sell compute, so when price of compute rises, they stand to gain the most (when contracts come up for renewal because more often than not they rent them out on 3 to 5 year contracts). Then you have giants like $IREN which started as a $BTC miner and now full into the AI HPC data center operations integrating vertically into the neocloud business. But the real question is: Why are rentals increasing? And why on old silicon? And to no one's surprise it's a demand and supply story, again. There's more demand, more compute needed, and price adapts Also, just because there's a new GPU like B200s, B300s and soon Vera Rubin, it doesn't mean that the old ones get thrown out of the window, especially if they are already colocated into data center, connected to the network, effectively cooled, powered, and actually working. Yes, at some point they will be replaced, but: - do you have the chips handy? are they in transit? were they even available at that size? - does it make economic sense when you factor in the replacement and all associated costs like installation, downtime, network, cooling, power and all? This is not a digital trading card that gets nerfed overnight with a software upgrade, we are in the physical world, which is a whole different ball game. Does this look like an AI bubble to you? And don't get me started on energy...

The quick depreciation trade is melting in front of the bears' eyes. Neoclouds will renew "old" compute at higher rates, while data center capacity is still constrained, providing a moat. And you are telling me these names deserve double digit drops when fundamentals keep getting better?