3 transcribed videos
2:53
This might be the hardest video I’ve made to date, despite the fact that I didn’t have to draw anything more complicated than a few boxes. In order to keep this under 3 minutes, I had to oversimplify A TON. I can’t caveat everything. But the bottom line is, if you want to put the key man risk of Jensen Huang — and the key company risk of NVIDIA — in context, you need to understand just how deeply they are invested across EVERYONE in the AI ecosystem. And while you’ll notice that there’s no real direct investment in the hyperscalers shown here, don’t be fooled. The frontier labs in whom NVIDIA has a massive investment are the ones renting compute and cloud space from the MSFTs, AMZNs, and GOOGs, of the world…not to mention the fact that investments like Marvell indirectly constitute an investment in their competitors’ custom chips. And despite their massive investment in OpenAI and other closed weight frontier model companies, they are deeply invested in the open weight models as well. Case in point? NOT PICTURED: their outright purchase of HuggingFace, which will show up in Q3. They bought the whole company at the nexus of the open weight world. These guys aren’t out there picking winners like the rest of us dummies. They’re invested in anything and everything that touches AI.
Nvidia is in bed with all its customers, not just the ones who are alleged Chinese spies. I set out to trace $1 of Nvidia's sales through their in...
27.5K1.4KNov 12, 2026
2:51
Your ULTIMATE guide to AI’s least understood business: the “Neoclouds”. Here are additional points: 1. The advance rate, meaning how much neoclouds can borrow relative to the amount of money they need to spend on Capex (to buy chips and equipment) can get QUITE high, in some cases 90-95%+ depending on the credit of the “off-taker”, so from Meta for example. 2. Asset backed financing and project financing (one of the groups I worked in at Morgan Stanley) isn’t new. What is unique about these GPU backed financings are the much shorter contract length; in the power world you often see power purchase agreements of 15-20 years and assets that have a much longer useful life (think a gas turbine). In a power deal, the debt is shorter than the PPA, and the PPA is shorter than the asset’s life, which leaves a long tail. Here the loan, the contract, and the asset useful life are much closer in length, so there is a shorter tail behind the debt. While Coreweave does have other types of debt, this structure accounts for over half and funded much of its massive growth over the past few years. They also have a decent amount of convertible debt, which is utilized by higher risk companies; I also worked in convertible debt origination at Morgan Stanley as well so let us know if you want a video on how that works. :) 3. And finally what suprised me was the fact that on the other side, so if you look at an Anthropic’s balance sheet, “take or pay” agreements to buy the compute do not show up as a liability on the balance sheet, rather just shown as a footnote!! While this is consistent with accounting rules, it’s surprising. With that said, there is a huge need for compute and these companies are major players in providing the needed capacity. Many are starting to own more of the infrastructure (Coreweave even tried to buy Core Scientific). #ai #Tech #aiinfrastructure #TikTokTechContest
When I heard Anthropic has agreements to spend $518 billion on compute, I had to understand what companies benefit from that half a trillion dolla...
12.8K640Nov 6, 2026
2:43
Here’s why NVIDIA is paying $13 billion to buy Hugging Face, the company OpenAI agents hacked into last month 👀
Why is Nvidia's acquisition of Hugging Face bad news for Anthropic and Open AI? Well, to understand, we have to go deep into Open vs. Close Weight...
16.2K837Nov 2, 2026