IA_817 retweeted
1/ Today we’re publishing the Databricks AI Index, showing trends from enterprises who manage AI with our Unity Gateway. Data covers >10,000 companies with broad representation (Digital Natives, Fortune 500, Global 2000, etc...). Enterprise AI consumption is changing quite a bit:
IA_817 retweeted
NEW: We continue to learn more and more about how Nvidia's plans to finance the AI boom were running into issues—even before Jensen was ousted.
Nvidia’s Days as an Open Checkbook for the AI Economy Are Numbered.
w/ @RWhelanWSJ & @KateClarkTweets for @WSJ
more tips? Signal me: 215-896-1307
wsj.com/tech/ai/nvidias-days…
IA_817 retweeted
Holy f*ck... Elon Musk's team just made hospital billing departments useless
my ER bill for a kidney stone came in at $11,480. I was about to sign up for a payment plan. my Grok Bot got it down to $2,900 in one phone call
Grok Bot makes phone calls. it sits on hold, talks to the billing rep and says up front it's an AI
what it did while I was at work:
→ sat through 23 minutes of hold music
→ asked for the itemized bill. they never send it unless you ask
→ found the same CT scan billed twice: -$3,200
→ I walked out after 3 hours. they billed a level 5 "critical" visit. got it recoded: -$2,440
→ $40 for a "mucus recovery system". it was a box of tissues. removed
→ asked for financial assistance screening. I qualified for 50% off
→ "pay in full today, get 10% off" it didn't pay. it asked me. I took the 0% payment plan
→ saved the reference # to my notes and set a reminder to check the new statement
31 minutes on the phone. $8,580 off. I didn't say a single word
all it took was 3 rules I sent it once:
1. on every call, say you're an AI calling for me
2. ask for everything I'm owed, nothing made up
3. anything that costs money: ask me first
the Grok Bot Research Desk setup is in my article. read it below ↓
IA_817 retweeted
What a game. This is why we college football.
Texas resiliency, and their defense. But remember LSU did throttle Clemson week 1 so they’re not out of the playoffs quite yet
IA_817 retweeted
Former Citadel head of equity quant research Dan Morillo on why AI can't generate alpha yet:
"It is trained on millions of documents, thousands of millions of documents."
"In finance, you would think of this as, it gives you the consensus."
"If you put a prompt in ChatGPT, what you get back is essentially some average of what a typical person might say."
"When you think about alpha generation, you don't make money off of the consensus. You make money off of differences."
"We just don't think it's at that stage at this point."
IA_817 retweeted
Recently spoke with a friend who works in L/S TMT at a pod shop. His team has integrated Claude fully into their workflow, including portfolio monitoring, tracking macro KPIs, and taking the first pass at building models and memos for your PM.
The last time he built a model from scratch was a few months ago. Ever since then, he has been plugging in his usual Excel template and key assumptions into Claude to get a quick and dirty model first. Then refine assumptions to create final analysis from there.
His biggest takeaway is that incorporating AI has opened up significantly more room to do more creative analysis that he had no time to focus on before.
Some portion of his time used to be spent plugging in numbers from a 10K/Q into his Excel or just downloading all the investor presentations from the IR site to track other disclosures and KPIs.
Now all of that has gotten automated with Claude. Instead, he has shifted more of his time to third-party data sources, speaking with consultants and experts, equity research analysts, and more management teams.
The biggest change has been the speed it takes to get caught up on a new name. If his PM came to him and asked to take a look at a stock he had never spent time on before, it would usually take a few days to a week to get fully caught up
Now it can be done in the span of a few hours or a couple of days at worst. The velocity of names that he is churning through has gone up significantly, and he is able to cover a wider array of verticals and sub-verticals as a result
He explicitly said his productivity has nearly 1.5x this time last year using the same exact tools. The difference is that the capabilities of these tools have improved so much that it feels like a night-and-day difference.
IA_817 retweeted
Steve Sarkisian is a freaking GENIUS!!! This play design just got @TexasFootball into the playoffs!
This is all play design, I’ll point out the details, just watch.
*Sorry I had to whisper I’m at dinner.*
IA_817 retweeted
neoclouds / hypers are financing vehicles for the cos selling tokens to the end user. their function is to bridge mismatch in cash flows between end user AI spend and infra investment.
so when they lean more on “customer prepayments”, this is essentially saying “we can no longer serve our function as financing vehicles.” it pushes the financing needs further down the chain, closer to needing to be met directly out of profits from end user AI spend / out of new equity capital from the investors in the companies selling tokens to end users (ie, VC).
it is AI infra’s version of a credit crunch, being passed onward from neocloud capital markets to the rest of the ecosystem.
IA_817 retweeted
Tae knows Nvidia better than anyone. Sound take:
"I don’t believe Nvidia’s story is over. The company has a deep bench of talent that does the work and I now believe it can continue to thrive under a new CEO. But it has to be the right leader. In my view, there are only two choices." @firstadopter
GPU rental rates aren't depreciating on schedule.
As of October 19, rental rates were above a six-year straight-line depreciation path from their first observed rent:
- A100s by 9%
- H100s by 54%
@mavolpi: "The demand is so high that even if the watt-to-dollar conversion is low, you're going to still use the asset"
Mike Volpi explains why older GPUs are holding their value:
"We thought this asset had a five-year life ... If you're in the business of financing an asset and that asset ends up having a longer useful life, then you're the beneficiary of that."
" In theory, at some point when you have sufficient Vera Rubins, it doesn't really make sense to run H100s anymore because it's just a worse way of converting electricity into dollars."
"The problem is right now that the demand is so high that even if the watt-to-dollar conversion is low, you're going to still use the asset."
"Ultimately the scarcity is in the manufacturing of the semiconductor itself... And so then scarcity persists, then the life of the asset continues to be long."
@mavolpi @RaghuRaghuram @appenz
IA_817 retweeted
Hello, offtaker expert here!! Apparently customer prepayments are now proof of extreme credit distress. Lmao. Someone should probably tell the entire infrastructure industry that advance payments are a financial emergency. We've apparently discovered a new law of economics on fintwit.
Customers prepay to secure scarce capacity, negotiate better economics, or help finance assets they intend to use. Sometimes because the supplier's alternative financing is expensive. Shocking stuff, I know. And btw, the 70% to 100% capex financing $CRWV discussed refers to asset backed loans. Customer prepayments are a seperate mechanism. Might wanna distinguish the two before declaring a new use as charity. We all know how companies love to do charity
Now, $CRWV absolutely has financing risks. Unsecured debt is expensive, interest costs are enormous. Nobody serious disputes that. The question is what the customer gets in return, what margins survive, and whether the contracts cover the capital costs. Not whether somebody PREPAID therefore CREDIT CRISIS.
IA_817 retweeted
Average discount on cheaper inference is 40.5% and all frontier models are available
Try it: CheaperInference.com
teapawt.net/CheaperInfer/status/21…
We combine customer demand to negotiate bulk discounts with AI labs and data centers.
500+ models available (including all frontier models)
ZDR available
Try us: CheaperInference.com
IA_817 retweeted
FireworksAI, TogetherAI, Baseten, Modal looking to buy their own GPUs and Datacenters.
This is all while Anthropic is max bid for LPS and TikTok is looking for 5GW of Colo.
At the end of the day, LPS will be constraint over software unless you’re the absolute best which is Anthropic and one or two of the inference providers.
Inference providers are increasingly looking to buy their own chips & run data centers themselves bc they can't get allocation. That's also pushed them to use dozens of smaller clouds, who don't always have the best security.
w/ @catherineperlo1
theinformation.com/articles/…
IA_817 retweeted
Replying to @GavinSBaker
Yes and this is why the neocloud credit crunch and the prepayment shift are the exact same story. The prepayment didn't kill the credit question for lenders…it changed who's answering it. The customers paying higher upfronts are now financing the buildout, and most of those customers are venture-funded inference companies paying deposits out of their last round.
That's why a 3-year B300 is cheaper than monthly right now, the discount is from the price of the customer's capital, not just the hardware alone.
Seeing a correction in the neo cloud market too (which is healthy)! Operators who can't fund a deposit lose the allocation but operators with real dock-to-live numbers get cheaper capital. The spread between these two is where we are at in the cycle and quite frankly this is normal for any commodity market growing (particularly one financed by credit).
The next step is lenders financing the end compute buyers against these contracts so they don't have to sell equity every three months to lock in cheap compute, and that's finally starting to happen now.
IA_817 retweeted
Living in SF is so crazy sometimes.
Woman posts on Reddit that a homeless man is repeatedly threatening her + her toddler at their bus stop - yelling at them and following them down the sidewalk.
Dozens of commenters tell her that SHE is the problem 🫠
IA_817 retweeted
RE: Nvidia. Over the last few days, I decided not to speculate over Jensen's future and to let the facts emerge before commenting.
Many people have cited my book as saying Nvidia would be completely lost without Jensen Huang.
My views have evolved in the several years since I wrote those passages.
I now believe there is a strong internal candidate with the technical expertise, experience, customer relationships and respect of employees needed to lead Nvidia effectively into its next chapter.
I will say more in due time.



































