the boot is too tasty for them

---

Gave up pussy, stopped doing toot Now you can't wait to give someone the boot.

---

I like the anti immigration stuff but she’s a typically boomer who hasn’t realized that times have changed and you can’t bull your boot straps up out of everything anymore.

---

VKTX still new boot goofin’

---

Using such rational thinking in an irrational market and when it’s Q4 to boot

---

OpenAI Needs More Than A Trillion Dollars ($500bn in operational expenses and at Least Another $625bn-$800bn+ for data centers), And There Is Not Enough Private and Venture Capital To Pay For It At no point is anyone asking how, exactly, OpenAI builds data centers to fill full of these GPUs. In fact, I am genuinely shocked (and a little disgusted!) by how poorly this story has been told. Let’s go point by point: $10 billion is not enough for OpenAI to build a data center: The 1.1GW Abilene Texas data center being built by Oracle and Crusoe for OpenAI required $15 billion in debt, and that’s not including the $40 billion of chips needed to power it, though I question whether all of them are going into Abilene (I’ll get to that later). In my premium article, I estimated that a 1GW data center costs $32.5bn and takes 2.5 years to complete. That might be generous. Matt Britzman, a Senior Equity Analyst with Hargreaves Landsdown says that each gigawatt of capacity generates $50bn of revenue for Nvidia. If correct, this would mean that a 1GW facility could cost perhaps as much as $75bn, or even more, when taking into account the other costs, like hardware, land, construction costs, permitting, the power infrastructure that said facility requires, and more. One thing that puts the figure closer to that which I estimated is the price of a proposed 1GW facility, co-funded by France and the UAE, which is expected to cost between $30bn and $50bn. Last week, Shirin Ghaffary of Bloomberg published an article that quoted unnamed OpenAI executives where they claimed a 1GW facility costs roughly $50bn, with $35bn going exclusively to Nvidia’s chips. Either way you cut it, $10bn is nowhere near enough. OpenAI Cannot Afford To Pay For Everything It Has Promised: OpenAI will burn $115 billion in the next 4 years according to The Information, but I believe the company intentionally leaked those numbers before the announcement of its $300 billion deal with Oracle, and the subsequent deal with Nvidia. In fact, OpenAI’s compute costs and promised data center expansion amount to more than a trillion dollars, with $750 billion of that mostly the cost of compute, meaning OpenAI will need actual real money to pay for it. If the $50bn-per-gigawatt figure cited by OpenAI’s execs is accurate, OpenAI will need to spend $500bn to create the 10GW of capacity required to unlock the promised $100bn from Nvidia. Even if the true figure is closer to my estimate of $35bn-per-gigawatt, OpenAI is on the hook for $325bn — or has to find partners to finance that construction. Either way, $325 is money that it does not have. That figure also includes OpenAI’s $19bn commitment to the Coreweave project, it planned $41bn in spending on Microsoft Azure by the end of 2028, the $22.4bn worth of business promised Coreweave, and the yet-unknown cost of renting Google’s compute, including that powered by its TPU chips. I also haven’t mentioned OpenAI’s other Stargate-related projects, including Stargate UK (1GW, so likely costing between $32.5bn and $50bn, or perhaps more), Stargate Norway (which has at least $6bn in GPUs), Stargate UAE (a 200MW facility that’ll scale to 1GW, so, again, anywhere between $32.5bn and $50bn), and Stargate India (another 1GW facility costing anywhere between $32.5 and $50bn). Although OpenAI has partners that are funding the upfront cost of those facilities (like Nscale, G25, Cisco, Nvidia, and Softbank), said funding likely is based on the assumption — or the requirement — that it actually uses that compute. It’s also unclear whether OpenAI’s partners will fund all of these developments, or just a significant chunk of them, with OpenAI on the hook for a chunk of it. Even if it’s just a small chunk, a small chunk of a (for the sake of argument) $50bn facility is still a lot of money. I also haven’t mentioned OpenAI’s planned $100bn spending spree on backup capacity. It’s unclear whether that figure is factored into the aforementioned spending, or something totally different, but based on what I’ve read, I’m leaning towards it being yet another cost that OpenAI almost certainly can’t bear. OpenAI Is Still Yet To Convert To A For-Profit, And Must Do So By End Of Year Or Will Lose $20 Billion In Funding From SoftBank: A few weeks ago, Microsoft and OpenAI co-published an announcement that they had “signed a non-binding memorandum of understanding for the next phase of their partnership,” which the media quickly took to mean a deal was done. No deal has been done. This is literally an announcement of nothing. Failing to convert will also see previous equity investments transform into loans that need to be paid back, with brutal interest rates to boot. Venture Capital Will Run Out At The Current Rate of Investment In Six Quarters, And OpenAI Needs More Investment Than Ever: The Information recently published some worrying data from venture capitalist Jon Sakoda — that “at today’s clip, the industry would run out of money in six quarters,” adding that the money wouldn’t run out until the end of 2028 if it wasn’t for Anthropic and OpenAI. In a very real sense, OpenAI threatens the future of available capital for the tech industry.

---