VCs love talking about the number of cofounders and how sometimes thereās too many..
@AnthropicAI x @databricks both with 7 cofounders showing that the best startups are truly outliers.
All the definitions of the ārightā way of identifying or building startups are wrong.
Thereās no ārightā way, thereās only the way that works for you!
@AnthropicAI x @databricks both with 7 cofounders showing that the best startups are truly outliers.
All the definitions of the ārightā way of identifying or building startups are wrong.
Thereās no ārightā way, thereās only the way that works for you!
14
4
135
9,062
Things I have a pretty decent idea about but I still have to go deep on since Iāll get 10 wildly different definitionsā¦
> long horizon
> out of distribution (esp data)
> intelligence per watt
> rl environments
> āforward deployedā
> revenue
> āfrontierā
> self improving
Sure, some of these are rightfully subjective, but youād be surprised at the variance of the responses š
> long horizon
> out of distribution (esp data)
> intelligence per watt
> rl environments
> āforward deployedā
> revenue
> āfrontierā
> self improving
Sure, some of these are rightfully subjective, but youād be surprised at the variance of the responses š
5
0
46
4,789
Founders, please ask your marketing team to exclude VCs if you decide to promote your product on X..
I get 3 DMs a day asking to "collab" on this "insane opportunity" which is *cough cough* a paid partnership.
Shows really poor judgement and word spreads around that you are to cue a big raise.
I get 3 DMs a day asking to "collab" on this "insane opportunity" which is *cough cough* a paid partnership.
Shows really poor judgement and word spreads around that you are to cue a big raise.
14
1
43
8,275
And definitely DO NOT follow it up with an email as well.. what is going on people?? š¤·āāļø
4
0
5
1,191
The main thing I wanted to dig into @AnthropicAI's S-1 was the compute contracts..
A few things that stood out (built in conjuction with Opus 5.5):
> At June 30, Anthropic had $265.6B of purchase commitments and $321.4B of leases that hadnāt started. Thatās a $587B subtotal, paying some conditional leases. The latest aggregate total is still blank in the filing.
> The contracts are very different. Google and Amazon require minimum spending regardless of usage. SpaceXAIās agreements have a potential value of $120.5B, but either party can cancel on 90 daysā notice after initial locked periods end in January and March 2027.
> Anthropic Direct separates chips, buildings, power and operations into different contracts. It lets Anthropic reserve a building early and commit to chips later. It also means one supplier can start billing before the others are ready. Equipment rent can begin before the compute is usable.
> Broadcomās $42B facility never hands Anthropic cash. Notes settle specific equipment rent bills. The facility excludes each equipment podās first quarterly payment and any rent due before it is ready for service. Anthropic has to find a way to fund those payments elsewhere.
> Those notes carry 9% interest, added to the balance quarterly, and can convert at 90% of the trailing 30-day average share price after IPO. New notes issued during specified revenue-breach periods get 13% interest and an 85% conversion price. The share-price measure is volume-weighted.
> That interest sits on top of the equipment leaseās own financing charge, whose rate is confidential. The investor chooses whether to convert the notes. Otherwise, principal and accumulated interest come due in cash.
> Anthropic treats Trainium, TPUs, NVIDIA and AMD GPUs as a shared fleet across model development, paying customers and internal use. It doesnāt disclose a fixed physical split between training and inference. The supplier bill can be fixed while the work assigned to the hardware changes.
> Revenue per dollar of total compute and infrastructure expense rose from $0.57 in Q2 2025 to $2.08 in Q2 2026. Pricing, utilization, technical improvements and when capacity arrives all affect that ratio. Itās a substantial improvement in monetizing the compute base.
> About 80% of H1 revenue was consumption-based. Most compute commitments however run for years. With quarterly releases, this bifurcation between usage and subscription will be interesting to monitor.
Source: sec.gov/Archives/edgarā¦
A few things that stood out (built in conjuction with Opus 5.5):
> At June 30, Anthropic had $265.6B of purchase commitments and $321.4B of leases that hadnāt started. Thatās a $587B subtotal, paying some conditional leases. The latest aggregate total is still blank in the filing.
> The contracts are very different. Google and Amazon require minimum spending regardless of usage. SpaceXAIās agreements have a potential value of $120.5B, but either party can cancel on 90 daysā notice after initial locked periods end in January and March 2027.
> Anthropic Direct separates chips, buildings, power and operations into different contracts. It lets Anthropic reserve a building early and commit to chips later. It also means one supplier can start billing before the others are ready. Equipment rent can begin before the compute is usable.
> Broadcomās $42B facility never hands Anthropic cash. Notes settle specific equipment rent bills. The facility excludes each equipment podās first quarterly payment and any rent due before it is ready for service. Anthropic has to find a way to fund those payments elsewhere.
> Those notes carry 9% interest, added to the balance quarterly, and can convert at 90% of the trailing 30-day average share price after IPO. New notes issued during specified revenue-breach periods get 13% interest and an 85% conversion price. The share-price measure is volume-weighted.
> That interest sits on top of the equipment leaseās own financing charge, whose rate is confidential. The investor chooses whether to convert the notes. Otherwise, principal and accumulated interest come due in cash.
> Anthropic treats Trainium, TPUs, NVIDIA and AMD GPUs as a shared fleet across model development, paying customers and internal use. It doesnāt disclose a fixed physical split between training and inference. The supplier bill can be fixed while the work assigned to the hardware changes.
> Revenue per dollar of total compute and infrastructure expense rose from $0.57 in Q2 2025 to $2.08 in Q2 2026. Pricing, utilization, technical improvements and when capacity arrives all affect that ratio. Itās a substantial improvement in monetizing the compute base.
> About 80% of H1 revenue was consumption-based. Most compute commitments however run for years. With quarterly releases, this bifurcation between usage and subscription will be interesting to monitor.
Source: sec.gov/Archives/edgarā¦
9
2
79
10,817
If you are a helpful angel that would invest in an infra series A and we donāt know each other, please holler!
1
0
7
750
This describes much more eloquently what I meant what evolving from context -> harnesses -> judgement means..
If you have built something in this space, and want to jam on product ideas, please hit me up š
If you have built something in this space, and want to jam on product ideas, please hit me up š
I think we are quickly evolving past needing surfaces to review intent/alignment and into needing better decision making "command centers."
As decision making across many contexts, which agents effortlessly own, is becoming the bottleneck.
This underscores the need for knowledgeable decision makers ā and less so about a bottleneck of us not understanding; better models are able to derive aligned intent and produce tasteful+correct solutions without too much steering. This enables the experienced and things are just starting to get fun as we babysit less and do less of "wait bro wtf did you just do?"
UI/UX:
Similar to intent/alignment surfaces, and perhaps ironically bc better decision making surfaces need to provide adequate context per decision (that is easy for a human to consume/understand).
e.g. similar things, but different:
A tool feels behind when it focuses on a side bar of message thread activity and inline html rendering vs a tool that manages multiple contexts, surfaces relevant html artifacts per thread, and abstracts the exhaustive message threads away from you.
9
0
38
5,448
Iām batting almost ~10% for a seed company that I have met that is eventually acquired in a few months from my first meeting..
The seed -> series A squeeze is at an all time high and is not going to abate anytime soon. Never been easier to dream, never been harder to scale.
New essay on this next week!
The seed -> series A squeeze is at an all time high and is not going to abate anytime soon. Never been easier to dream, never been harder to scale.
New essay on this next week!
I counted more M&A happening than deals this week..
Itās only going to increase as seed companies with limited runway face a choice: strap in or join a hyper growth company.
Itāll be interesting to see how the landscape shifts for the rest of the year.
6
1
92
10,479
If you are curious about how out of distribution this El NiƱo season is, two charts give a pretty clear picture..
3
0
61
11,110
