Tail Risk

one undersea fiber cable laying company got 100x booking (2025 vs 2024) from hyperscalers to lock up capacity for the next 10-20 years..

The most interesting comment I read in today’s report: All of Anthropic's unusual moves around the GPT-5.5 launch ultimately converge on a single conclusion: in order to secure compute, Anthropic must bind itself far more deeply — and far more dependently — to those who possess these physical resources. This conclusion was crystallized in the $100B deal formalized a few days ago. The most representative example is the deal with Amazon. The key terms are as follows: - Amazon will inject an additional $5B into Anthropic immediately, with up to $20B more in follow-on investment to come. Combined with the existing $8B, cumulative investment reaches up to $33B. - Up to 5GW of compute capacity secured, of which roughly 1GW is scheduled to come online by year-end 2026. - Anthropic has committed to spending $100B+ on AWS technology over the next 10 years (including usage requirements for AWS silicon such as Trainium and Graviton). - Purchase rights secured for Trainium2/3/4 and future Trainium generations. Trainium2 is currently sold out, and Trainium3 is largely booked as well. - The Claude Platform will be offered natively on AWS. Just how large is the 5GW Anthropic has committed to? It is equivalent to roughly five nuclear power plants. Given that Microsoft's total global data center footprint in 2024 is estimated at around 5–6GW, this means Anthropic alone is locking in incremental capacity — for AI training and serving — that rivals the entirety of MSFT's historical physical infrastructure. Coupled with Anthropic's announcement of having reached $30B ARR, the market is reading this deal as "Anthropic pre-signing Amazon's invoice in order to keep its growth going." It is also worth noting Dario Amodei's (Anthropic CEO) remarks accompanying the announcement: "Users tell us that Claude is becoming increasingly essential to how they work. We need to build infrastructure to keep up with rapidly growing demand." Last month's blog post from Anthropic — directly admitting that "surging enterprise/developer demand is placing unavoidable strain on our infrastructure, affecting reliability and performance" — was effectively a teaser for this deal. What matters is that the structure of this deal is more favorable to Amazon than to Anthropic. Amazon has already invested in OpenAI as well (up to $50B). In other words, the more fiercely OpenAI and Anthropic compete to eat each other's lunch, the more Amazon benefits simultaneously along three axes: cloud usage fees from both, adoption rates for its in-house silicon (Trainium as XPU, Graviton as CPU), and visibility on the recovery of data center CAPEX. This is structurally almost identical to the valuation premium Google historically enjoyed under the "full-stack player" framing. The market, however, has not yet fully come around to recognizing Amazon in this light. (Sentiment has admittedly improved compared to two weeks ago.) Amazon is still trading near a 10-year low on CY26 EV/EBITDA. But as the rivalry between the two AI labs escalates and Amazon begins to collect its "toll-gate revenue," we believe the market is likely to gradually move toward a re-rating. (Excerpt from Mirae Asset Securities’ AI Hot Issue report, dated April 24, 2026) $AMZN

wouldn't it make sense for TSLA/SPACEX/XAI to just acquire intel?

*ANTHROPIC’S MYTHOS MODEL ACCESSED BY UNAUTHORIZED USERS

$GOOGL and $AMZN, among others, are becoming serious CHIP companies that investors are not giving enough credit for. Right now, $AMZN has around 1-1.5 GW of Trainium chips while $GOOGL has 2-3 GW of TPUs running. Given the OpenAI and Anthropic deals announced for the next few years, both companies will sell more than 12 GW of Tranium and TPUs. When you add in internal demand from $GOOGL and $META, we are looking at +15 GW of ASIC chip sales. If those chips were sold directly, that would be +$500B in chip revenue. $AMZN already said that with Graviton and Trainium, they are running a $50B annual revenue chip business, growing at triple-digit YoY rates. $GOOGL's TPU and $AMZN's chip business would each generate more revenue than $AMD does today ($34B). $AMD trades at a $450B market cap.

is anyone worried about mythos opening the floodgate for enabling hacks across defi? I'm concerned and I want to pull everything. If you aren't worried, find a way to short defi?