Oxford Analytics

SpaceX LOST $4.9 billion in 2025 on revenues of just $18.7 billion. Anyone buying into the wildly inflated IPO at a valuation of nearly $1 trillion is a complete idiot. Musk is a charlatan who plans to merge $TSLA, with its falling sales, into SpaceX after the IPO. IPO buyers are Musk’s exit liquidity.

Some consumer stocks (both consumer staples & consumer discretionary) that were in the green during today’s AI-linked selloff: $MCD $YUM $DPZ $KO $SBUX $WMT $PM $MO

Copper mining stocks got hammered today along with semiconductor and all AI infrastructure stocks: $SCCO -10.9% $FCX -9.1% $TECK -8.3%

Gold is down 3.4% today illustrating that it’s not a hedge against equity market volatility. On the same day semiconductor stocks are selling off sharply along with AI infrastructure stocks in general both gold and silver are down. $GLD $SLV $XAU

Anyone buying the SpaceX IPO is either incredibly gullible or incredibly cynical. Nothing in between. The fantasy revenue projections by the underwriters Goldman Sachs & Morgan Stanley are entirely worthless. The equity analysts putting their name to such drivel should be barred from the industry just as much as Andrew Left of Citron. SpaceX at $1.75 trillion or 100x sales is absurd. It isn’t even worth half of that.

@CommodMkt @AlanEyre1 Trump will make a face saving deal with Iran before we reach “tank bottoms.” This is alarmist nonsense clickbait. There’s no chance of the US running out of oil given its now the largest EXPORTER in the world when you add in refined products.

Thanks @AlanEyre1. From Washington to Beijing, the response has been the same: reassure markets with words and hundreds of millions of barrels from strategic reserves. The gamble is that governments can bridge the gap until supply returns and prices fall. The problem is that a falling oil price is not evidence of rebalancing. It is evidence that we are consuming the buffer and mistaking it for abundance. There is a difference between a market clearing and a market solving. What we are witnessing is destocking: the drawdown of finite inventories accumulated over decades. You can only do that once. The article notes that “inventories and government reserves run low”, but that understates the issue. Strategic reserves buy time, not supply. When tank bottoms arrive (likely later this summer), there is no price signal that conjures a new barrel into existence. The “stream finds its way around the log” metaphor is seductive but flawed. It assumes the total volume of water is unchanged and that we don’t mind being thirsty. The Strait’s closure has not simply redirected supply; it has removed supply while the world burns through stored reserves to cover the gap. When Jimmy Carter asked Americans to turn down the thermostat, he was punished for acknowledging physical constraints. His successors learned to promise abundance instead. That lesson is being applied again today, and the consequences will be proportionally larger for the delay.   You cannot print molecules. And you cannot destock your way to energy security.

Let's revisit this thesis when we stop tapping our reserves and begin to hit tank bottoms. What do you think, @CommodMkt ? nytimes.com/2026/06/04/opi…

Mining stocks are selling off right along with semiconductor stocks this morning. There’s no diversification benefit from owning mining stocks - particularly copper mining stocks. Copper is tied up with datacenter demand just as chips are. $SCCO $FCX $MU $NVDA $AVGO

The head of $MSFT AI said Anthropic’s Claude Code is too expensive for Microsoft’s own employees. Anthropic & OpenAI are wildly overvalued. Their revenue projections are completely unrealistic. Corporate customers will switch to cheaper AI models. Avoid the IPO’s. Ignore the hype.

OpenAI confirming that the cost of using AI is much too high. Token costs to complete simple tasks are excessive. Companies are capping AI spend. OpenAI & Anthropic’s revenue trajectories won’t be as expected. Don’t buy the IPO’s.

$UBER blew its entire 2026 AI budget by April and is now capping employee spend on tokens. They’re a canary in the coal mine. AI compute isn’t cost effective now that the labs have switched from flat rates to usage based billing. Burning through millions of tokens to complete tasks that could be done by entry level workers doesn’t make any sense. That’s why the AI labs, OpenAI & Anthropic, are rushing to go public. They’re seeing the cost problem up close with their customers. Don’t touch the IPO’s. Ignore the hype.

Semiconductor stocks are getting hammered this morning after $AVGO results that guided towards AI demand that was just in line with expectations. Goes to show how much wild optimism is baked into the stock prices of the semiconductors. $SOXX $SMH $DRAM

The AI Bubble is being driven by the fantasy that AI can replace human labour cost effectively. It can’t. The cost of compute is too high. Semiconductors are too expensive. GPU’s, CPU’s, even a commodity like NAND flash memory. Power is too expensive. Water is too scarce. Key metals are too expensive. It’s cheaper for humans to do the work than burning tokens using Claude Code or OpenAI Codex. The layoffs were premature. Humans are cheaper than compute.

The top 5 PM’s at CPPIB earn more than $5 million each in total compensation. I guarantee none of them have outperformed $QQQ / $SPY / $XIU over any interval. Canadians are being ripped off by clowns with CFA’s who know they can’t beat the markets but will charge you a fortune for their efforts. @cppib @acoyne @PierrePoilievre @JonErlichman @rcarrick @EconguyRosie

You’re either positioning yourself defensively for the AI bubble to pop or you’re aggressively chasing semiconductor stocks higher with leveraged ETF’s. Those are the two extremes of the market right now. Some large hedge funds have AI exposure through the hyperscalers but no semiconductor holdings. Christoper Hohn’s TCI Fund owns $GOOGL & $MSFT. Bill Ackman at Pershing Square owns $AMZN, $MSFT, and $META. The biggest multi-manager HF’s, Citadel, Millenium, Point72, Balyasny all have significant positions in both the semiconductors and the hyperscalers. I wouldn’t touch any of the semi stocks at current levels. Their earnings are at cyclical highs and unsustainable long term. Particularly the memory stocks like $SNDK, $MU, Samsung Electronics & SK Hynix.

Don’t tell me we’re not in an AI bubble when Goldman Sachs’ Non-Profitable Tech Index is up 200% over the last year and is reaching its 2021 Meme Stock Bubble peak.

NAND flash memory is a commodity product. There’s a shortage right now but it’s a cyclical shortage not a structural one. Once the shortage disappears there’s going to be a hard landing for the shares of $MU, $SNDK, Samsung Electronics and SK Hynix and $EWY.

A wall of money has exited crypto and entered semiconductor stocks over the last month. Crypto bros are now day trading chips stocks. bitcoin:native $MU $SNDK $MRVL $AVGO $NVDA $INTC $AMD

Corporate spending caps on AI will wreck any forecasts for profitability of the AI labs - OpenAI & Anthropic.

All the crypto bros have piled into semiconductor and memory stocks. bitcoin:native $NVDA $MU $MRVL

Roundhill threw up 15 different thematic ETF’s and 2 have crossed the billion dollar mark of AUM: $DRAM - $15 billion $MAGS - $4.6 billion

A lot of supply will be hitting US stock markets this year. Not just from the IPO’s of SpaceX, Anthropic, and OpenAI but also from the expiration of lockups.