_sword

This is a silly article. Since MSFT took a ~49% stake in OpenAI, it records its share of OpenAI's net losses in the other income line under the equity method of accounting. MSFT is offsetting its taxable income based on a prior investment

I was modeling configurations purpose-built for running specific models in specific workloads. I was trying to figure out how much of a gross margin drag some software companies could have if they hosted their own models and served them up as APIs or as integrated copilots with their other offerings

I spoke with management at a couple companies that were training models, and some of them expensed the model training in-period as R&D. That's why

I've done the modeling on this a few times and I always get to a place where inference can run at 50%+ gross margins, depending mostly on GPU depreciation and how good the host is at optimizing utilization. The challenge for the margins is whether or not you consider model training costs as part of the calculation. If model training isn't capitalized + amortized, margins are great. If they are amortized and need to be considered... yikes

Neat, more scalable intelligence for me to tell "plz fix" over my code

Sounds like meta gets the data and also anything they can scrape together about what the other labs have been doing to get better results than meta

That's why they so despise each-other; they're the same

The self dealing king Sam Altman strikes again

Even before LLMs were popularized, the shift to remote work made hiring awful in my experience. In finance roles, I had candidates who aced their tests and projects but then showed up to the job unable to competently use excel or write coherent sentences in English. Phone / zoom interviews all went fine, but clearly there was rampant cheating during remote projects.

Could you name any products?

Makes sense and was only a matter of time considering it has essentially no revenue growth to date this year and Non-GAAP margins in the low-to-mid 30's %. With near-0% revenue growth, investors will expect a SaaS company to post 40%+ margins.

Reads like a play to dilute the power of executives that departed

OpenAI is definitely planning to IPO within the next 12-18 months after it restructures into a for profit.

CEG jumping 20% on the MSFT deal means it isn’t all priced in imo. Everyone has known there are rising energy needs and had bid up CEG. Then the Three Mile Island restart news hit and apparently it wasn’t already priced in

That’s the whole point of Forbes today

There’s some research here that the NYTimes cited in an article on the topic of TikTok deplatforming certain topics

https://networkcontagion.us/wp-content/uploads/A-Tik-Tok-ing...

I thought wind at first because it's been stormy lately

That was eerie! I felt my apartment building start to rock and sway as the walls were creaking. Not something I’m used to as a NY native, and not something I loved to experience 20+ floors above the ground.

At virtually all banks, equities analysts are banned from trading in their coverage. Banned as in, if you, your spouse, your dependents, etc. have an interest you didn't proactively disclose and dispense with you're fired on the spot.

FINRA regulation states that registered equities analysts (i.e. the ones working at banks) at a minimum cannot trade against their ratings [0]. At most / all banks there are further restrictions that ban trading in coverage.

[0] https://www.finra.org/rules-guidance/rulebooks/finra-rules/2...

I took undergrad quantum physics with Prof Greene and it was a great time. I definitely could have chosen a better professor for the math, but the theory and his banter was fun.

The tech could be really cool if e.g. classifiers could be represented within the probability space modeled on their hardware. However their shaman-speak isn't confidence inducing.

At this point I wonder how much of the GPT-4 advantage has been OpenAI's pre-training data advantage vs. fundamental advancements in theory or engineering. Has OpenAI mastered deep nuances others are missing? Or is their data set large enough that most test-cases are already a sub-set of their pre-training data?

The White House doesn’t differentiate! If you’re foreign and training bigger models on US cloud infra, you’re reported

https://www.whitehouse.gov/briefing-room/statements-releases...

The White House’s AI regulatory order requires big cloud providers to monitor for and report foreign clients (ie Chinese) for using their services to train AI fyi. Monitoring for malicious activity isn’t nearly as invasive as what the government is requiring

Like a Windows-focused copy of the the Self Operating Computer Framework. There's clearly something here with this approach if it's popping up in multiple open source frameworks

I wonder if he's pitching fusion power and AI to the middle east as a post-peak oil hedge for their economies

He's probably trying to build out a vertically integrated AI conglomerate spanning power, chip fabs, software, etc. And he's great at marketing in the meantime

Is this because of actual energy needs? Or because Sam is trying to drive investments into the fusion energy company he's backing?