i like golf and systems @voltcapital | prev @Delphi_Digital
Joined June 2011
- Tweets11.6K
- Following599
- Followers6.1K
- Likes14.6K
Rob retweeted
live thinking: spice banks
- who actually wants to bank the businesses buying compute hours? most prefer assets, but what about flows. finance purchases across suppliers, understand what is earns via capacity, commitments.
- take inference business collecting from customers / paying suppliers on a slew of schedules. how do you get comfortable to lend against the cash left over?
- customer has cash sitting w/ supplier before token gets served. finance portion of it. get access to collections, margins, contract expiries. small wedge into much larger purchasing decision. you now know 1) when they need more compute 2) what they paid 3) whether they used it wisely.
- opens door to matching supply&demand. this is where a lot of people comp to marc/glencore. renewals come up, bring suppliers, negotiate terms, finance the purchase.
- keep finding things worth financing because youre already on the inside. starts to look like a merchant bank. layer on a bunch of products.
Rob retweeted
I am excited to be an investor in trade(xyz) as they partner with ICE to bring 24/7 perpetual futures on Hyperliquid to the United States. As a market participant, I am looking forward to to using their innovative product to enhance my investment process.
Rob retweeted
We are building atlases of the intangible economy.
Neuroscience provides some good intuition here.
It’s often the case that building atlases is simply impossible without the right neurotechnology, and it makes sense to co-design the atlas and the technology to acquire (or generate) that atlas.
For example, the Enigma project at Stanford, which is building the largest functional atlas ever collected, was empowered by the creation of next-generation Neuropixels, increasing throughput by a factor of 4.
A holistic strategy in the age of abudant intelligence optimizes the creation of new tools vs. using existing tools to build atlases of the intangible economy.
Simply, what is even more valuable than a static dataset is a data generation capability (both the right to and the ability to generate high signal data). This is how we engineer economically valuable serendipity.
The firm itself becomes a scientific instrument.
few in data are talking about owning differentiated sensors to capture novel data - everyone is red queen racing to unique sources of data (enterprises, bio, etc.)
but by controlling the sensor in which a certain form of data can even be made legible, you can control all downstream data feeds (ex: halter and agriculture)
ex: in the limit this looks like space telescopes (high capex, R&D intensity, regulatory permission, etc)
science.nasa.gov/mission/rom…
a few friends have begun to monopolize the most esoteric of sensors and are exhibiting the extreme form of pole position
extreme pole position (read: MONOPOLY) involves controlling the interface where reality is made legible to intelligence
Rob retweeted
We wanted to better understand what AI safety looks like in practice, so we brought @zachary_horvitz on @TheDryRunPod to walk us through his research on long-tail model behavior.
Some takeaways:
1. Rare model outputs aren’t always harmful. Exploring many possible answers can also help uncover useful discoveries. Zach connected this to how AI is helping researchers explore problems in math and biology.
2. A model refusing a harmful question once doesn’t mean it will refuse every time. As models become safer and more useful, more people use them. Even a small chance of harm matters when that usage reaches millions of people.
3. Finding rare harmful outputs is only part of the challenge. Researchers need ways to estimate their likelihood without generating thousands of answers for every question. They also need to judge those answers, often using other AI models that can make mistakes too.
We learned a lot from Zach and we’ll be exploring more of these questions on the pod.
What does rare AI behavior mean for safety?
Zach Horvitz (@zachary_horvitz), a PhD candidate at Columbia researching rare behavior in generative models, joins Dry Run to explain what rare harmful behavior looks like, how researchers measure it, and what it means for the safety of models people rely on.
00:53 What AI safety looks like today
04:05 What rate of harm is "acceptable"?
06:56 Why the same prompt gives different answers
08:08 Why rare harmful outputs matter at scale
11:36 Asking the same harmful question 10,000 times
12:13 How to estimate the odds of a harmful answer
13:50 Using AI models to judge other models
16:28 How human feedback rewards flattery
18:42 Why safer AI also gets used more
24:19 Testing models without access to their internals
27:04 How rare outputs could drive scientific discovery
Rob retweeted
had fun going through all my favorite financial history books and putting this together
as a wise man once said, alpha is hidden in the texts that many are too lazy to read
Our report “The House Always Wins: Why Speculation is a Theme of the Decade” is out now.
Speculation runs rampant when new tech rearranges what’s possible or the system stops working for most. AI and crypto are now fueling both.
Read it for free here.
delphi.link/Speculation
Rob retweeted
Alex is right here. Don't trust the eval industrial complex.
Most public benchmarks are terrible, and top rank there is a negative signal. If you're building products, you need to create your own private evals.
The private evals need to deeply correspond to what your users want and the unique knowledge and know-how you have in the domain. You also need to avoid open-source code as much as possible.
META’S HEAD OF AI SLAMS AI BENCHMARKS
"it’s some PhD student who made these benchmarks"
"you’re kind of outsourcing all of your product thinking to this viral benchmark"
"so clearly that’s wrong"
I think there is something to this. the whole industry developed so rapidly that there was hardly any time to establish proper benchmarking. companies used what was available, competitors then had to catch up on the same benchmarks, and some of them eventually became industry standards
apparently there is quite a lot of randomness and chance in how we are measuring AI progress today
Rob retweeted
big one. we teamed up with Matera and @solana to bring X9.150 into business payments
X9.150 is the first US national standard for QR payments, built by 70+ orgs, and the biggest US banks are lining up behind it
we're putting it to work in @altitude bill pay: signed recipient details on the invoice, verified through the X9.150 standard, settled in usdc on solana with the invoice that marks itself paid on the recipient's end
catch live demo at money20/20 this week
Find us at Money20/20 USA this week for a demo of Altitude Bill Pay using the new ANSI X9.150 QR payment standard.
See signed payment instructions, USDC settlement on Solana, and payment status updates in action.
Learn more: altitude.xyz/blog/raising-th…
Rob retweeted
agents are getting smarter and everybody will have an assistant helping them run their business
these agents are becoming the interface for how we use our tools. your assistant should be able to work with your finance stack using the context it already has about your business
@altitude MCP is live and this is our first step towards making Altitude the business account your agent loves
Ok @bot, analyze my entire transaction history and figure out how to turn this into a trillion dollar company. Make no mistakes.
Say hello to Altitude MCP v1.
Connect your AI agent to Altitude and put your transaction data to work.
Get started: docs.altitude.xyz/mcp/overvi…
Rob retweeted
we just bumped @altitude earn to 3.5% APY
the altitude earn setup is what we wished we had at squads all these years:
— backed 1:1 by short-term U.S. Treasuries custodied at BlackRock
— no minimums, no lockups, T+1 liquidity
— fund it with stablecoins or fiat
— powered by stablecoins and formally verified code running on the fast chain
— available in 100+ countries
businesses can open an Altitude account today, complete verification and move funds into Earn
rewards accrue daily and are paid monthly. withdraw back to your main account at T+1 when it's time to deploy capital
built for onchain teams managing stablecoin treasuries and global businesses that want to earn on their dollars
app.altitude.xyz/signup
Rob retweeted
Paperwork no longer needs to be a side quest.
Download account statements, balance summaries, account details, card statements, transfer confirmations, and more directly from Altitude.
Learn more: altitude.xyz/blog/financial-…
Rob retweeted
Better have a better long term business than selling data to labs
6-12 months ago the large labs started buying bio data
this spend is starting to ramp very aggressively. the labs are somewhat indiscriminate buyers with enormous budgets. the therapeutics market has long had buyers of data (large and small pharma) but they've historically been the exact opposite of huge-budget and indiscriminate
this shift will distort the bio startup market in a bunch of ways. there will be a lot of short-termist behavior to try to get in front of this capital firehouse. but unlike the AI data labelers (mercor, et al) and the robotics labelers (mecka, et al), the bio startups serving up this data have a chance to emerge from this period with businesses that don't rely on selling data
the best bio startups will use these sudden resources the labs are dumping on them to supercharge their novel assay and work towards independently powerful models and eventually directly produce therapies
bio is the final frontier and its a very exciting time ahead for bio startups IMO
Rob retweeted
Just like with oil, the market outside the majors is the true marginal market for compute.
Rob retweeted
SPVs have become a commonly used structure, especially in AI financing.
The valid uses are well understood, such as letting a lender take exposure to one slice of a business, keeping a failure siloed, and managing capital rules that don't make sense.
They become questionable when used to keep obligations off the balance sheet, which contributed to Enron's downfall, or when unregulated companies keep reaching for them.
@austincampbell from @ZKZeroKnowledge
How is the AI buildout financed?
Austin Campbell, founder of Zero Knowledge and adjunct professor at NYU Stern, dives into the ins and outs of the AI buildout, especially how it is financed, who's doing the financing, the role of debt, and how the AI boom compares to past bubbles.
03:39 Components of the physical buildout
07:41 Capex spent so far vs. future commitments
10:13 Why Big Tech buybacks are dropping
13:18 How SPVs are structured: good vs. bad uses
20:13 AI makes money, but can it pay back the capex?
20:59 Who takes the hit if demand falls short
24:33 How today's SPVs compare to 2008
32:25 Minsky's three stages of a bubble, applied to AI
Rob retweeted
cant miss. hosting a small dinner on 10/8 for those staying the week in LA
.@DiscipulusVent Fall Cohort Starts Today in El Segundo
Build for America 🇺🇸
More coming soon...
Rob retweeted
How is the AI buildout financed?
Austin Campbell, founder of Zero Knowledge and adjunct professor at NYU Stern, dives into the ins and outs of the AI buildout, especially how it is financed, who's doing the financing, the role of debt, and how the AI boom compares to past bubbles.
03:39 Components of the physical buildout
07:41 Capex spent so far vs. future commitments
10:13 Why Big Tech buybacks are dropping
13:18 How SPVs are structured: good vs. bad uses
20:13 AI makes money, but can it pay back the capex?
20:59 Who takes the hit if demand falls short
24:33 How today's SPVs compare to 2008
32:25 Minsky's three stages of a bubble, applied to AI
Rob retweeted
i hear (frequently) from people who try to over-intellectualize reindustrialization through various lenses in finance, economics, labor, technology, "free" market competition, etc. it's quite simple and doesn't require hardly any justification at this point. let me paint you a picture: the final crescendo of super intelligence is total mastery of atoms, and a semi-likely multipolar world of superpowers. the intermediate step that's missing right now is to build as much net-new tech-enabled capacity as possible. the transfer of tribal and process knowledge into compute is critical and very similar to how we've learned to control self-driving cars and robots. this is not something even the smartest model can predict/simulate with the level of necessary precision for real production, and these are glaring open deficiencies in all frontier models until they have data (live factories) to feast on. that learning rate will increase vastly as a function of overall tech-enabled capacity, and potentially displace frontier labs because they lack production experience and actual capacity themselves once intelligence is commoditized. the faster we do all of this the faster we create the potential for actual abundance.
