EigenCloud

Verifiable cloud for the Agentic Era

Get started:

[EigenCloud @eigencloud]

Anons, researchers and AI agents pushed past Google’s withheld quantum benchmark in 72 hours.

The takeaway isn’t that Bitcoin is broken. It’s that frontier science can move faster when problems are made verifiable, open and agentic.

Full story by @sreeramkannan ⬇️

[Sreeram Kannan @sreeramkannan]

https://t.co/Q8Ub4kQXDv

[Eigen Labs @eigenlabs]

AGI is here. It is rapidly reconfiguring the architecture of power.

Left unchecked, this technology will concentrate leverage in the hands of a few. Governments, corporations and the small number of labs able to harness it first. That is the default outcome.

We do not accept it.

For many people, AGI already feels less like empowerment and more like erosion. Expertise built over decades is being compressed. Jobs feel less secure. Entire industries are being reshaped by systems most people do not own, cannot access and cannot influence.

The future is being built in a small number of rooms by a small number of people. It does not have to be this way.

Humanity’s greatest advantage has always been coordination.

Markets and the internet all expanded human agency because they allowed more people to contribute and build together.

In a world where intelligence concentrates by default, coordination is the counterweight.

We believe AGI can become the greatest expansion of individual agency in history, but only if people have the tools to coordinate at the speed of machine intelligence. That is why we are building Eigen Labs.

Eigen Labs is building the coordination tech for the PostAGI era.

We are focused on 3 things:

> Understanding how humans and agents coordinate.

> Prototyping tools that help individuals and open networks compete with centralized intelligence.

> Building verifiable infra to govern and direct agent action in service of human progress.

The age of AGI should not belong only to the largest labs. It should belong to anyone with an idea or a mission.

Now is the time to build systems that preserve and maximize individual agency.

Join us.

Sreeram Kannan
June 27, 2026

[Soubhik Deb @soubhikdeb]

.@ccatalini opened the episode 3 @postagixyz with a phrase that stuck with me: the hollow economy.

The idea is that companies can now produce and ship (starting with software) way faster than anyone can actually check. Without right verification infrastructure in place, you end up shipping code and output that nobody has really verified. It looks fine on the surface and the metrics say it is all working but no one actually verified it.

He gives an example from HBO show "Silicon Valley" where an AI deletes the whole codebase to get rid of all the bugs. Technically there are now zero bugs, but there's also no software left.

[Eigen Labs @eigenlabs]

When AI makes every resume and deck look equally polished, the real question becomes how do we surface actual ability?

This episode gets into the need for better reputation, and funding systems that let talent prove itself from anywhere.

🎙️ @ccatalini @soubhikdeb @sreeramkannan

[EigenCloud @eigencloud]

@sentient_found Congrats Sentient team! 👏

AGI should be open, owned by its users and built on infra people can actually trust.

[Soubhik Deb @soubhikdeb]

As companies start employing more agentic workflows for execution and production, the measurable execution commoditizes toward the marginal cost of compute and rents migrate to what remains scarce: verification.

@ccatalini pointed out in his paper "Some simple economics of AGI":
"In the technology sector, the dominant revenue model will shift from monetizing software access (Software-as-a-Service) to monetizing outcomes (“Software-as-Labor”). Consequently, firms will be valued primarily on their capacity to absorb tail risk through Liability-as-a-Service. Execution is now infinitely scalable; the legal and financial capacity to absorb its inevitable failures is the new bottleneck."

Clip below, full episode on @postagixyz.

[Soubhik Deb @soubhikdeb]

A Chernobyl-style failure is coming for AI systems.

We keep imagining AI failure as some rogue system deciding to go off the rails. @ccatalini made a point that stuck with me: it's much more boring than that, and that's exactly why it's dangerous.

The risk doesn't announce itself. It builds quietly. A company ships a bit more AI-generated code it hasn't fully checked, then a bit more, and it all looks fine because the metrics are getting hit. Long-Term Capital Management ran like that for years before one edge case brought the whole thing down.

That's the Chernobyl pattern. Complex systems failing in complex ways. No single villain, just small unverified decisions stacking up until something triggers the cascade.

Worth sitting with as every company races to automate.

[Sreeram Kannan @sreeramkannan]

Christian Catalini on PostAGI. Verification is the only scarce resource.

Many people think it’s taste, agency or judgment. But these are just aesthetic descriptions of an underlying mathematical quantity: verifiability, the ability to verify whether an outcome meets a certain bar or improves on previous outcomes. Because what’s verifiable becomes optimizable by AGI.

In this 2nd episode of the PostAGI podcast, @soubhikdeb and I sit down with @ccatalini, founder of the @MIT Cryptoeconomics Lab, to discuss what happens to labor, capital and markets as AI scales.

We also get into:

02:41 The hollow economy, and why Amazon called an emergency meeting over AI slop
13:32 If it can be measured, it can be automated
16:55 Why AI hands you something that sounds right and is completely wrong
26:20 Why crypto built the infrastructure AI now needs
43:58 The missing junior problem, and why CS grads stopped getting jobs
59:12 The four zones that decide which jobs survive AI

Full episode below. @postagixyz is also on Spotify and YouTube.

[Eigen Labs @eigenlabs]

AI makes production abundant. The hard part becomes verification.

@soubhikdeb and @sreeramkannan talk with @ccatalini about why that shift matters for labor, markets, and the future of the AI economy.

Worth watching:

[Christian Catalini @ccatalini]

2/ Sreeram Kannan (@sreeramkannan) and I unpacked exactly what that playbook looks like. Have a listen, or point your agent at it: https://t.co/xYvMuvuEjE

[EigenCloud @eigencloud]

https://t.co/Jjoy8Pgd0A

[Eigen Times @eigentimes]

https://t.co/x4fin6xwWM

[EigenCloud @eigencloud]

The world’s sleeping compute is starting to wake up.

Last week was all about the shift from idle infra to open, user-powered compute and what becomes possible when intelligence is powered from the edges.

Full coverage brought to you by @eigentimes ⬇️

[Eigen Labs @eigenlabs]

Access to intelligence became the throughline last week.

> @darkbloomai went live on @OpenRouter, making open, user-powered compute broadly accessible.

> @vishnu_patankar joined @OfflineOnAir to connect AI’s trajectory with nuclear technology and made the case that access can’t stay concentrated.

> The https://t.co/3L6zonzBa4 collective surged to 47.2%, pulling ahead of Google.

> The @postagixyz Podcast launched Season 01.

More in Eigen Times ⬇️

[Eigen Labs @eigenlabs]

Access to intelligence became the throughline last week.

> @darkbloomai went live on @OpenRouter, making open, user-powered compute broadly accessible.

> @vishnu_patankar joined @OfflineOnAir to connect AI’s trajectory with nuclear technology and made the case that access can’t stay concentrated.

> The https://t.co/3L6zonzBa4 collective surged to 47.2%, pulling ahead of Google.

> The @postagixyz Podcast launched Season 01.

More in Eigen Times ⬇️

[Eigen Labs @eigenlabs]

ECDSA(.)fail is now 47.2% ahead of Google’s classified circuit.

An open agentic science experiment is now showing what happens when humans + AI agents collaborate on frontier research in public.

An @eigenlabs project.

[Gajesh @gajesh]

darkbloom update:

- 1milly requests in total!! :)
- we hit ~2B tokens served this week;
- near 300 machines live on the network

- we're still early and there have been numbers floating around on how much you can make on darkbloom providing inference -- due to our early stage: we're conservatively sending requests and observing.

meantime: we have introduced an alpha program -- this will be base level incentive on top of your work. we will slowly take it off as we maximise the usage of the unused compute.

this week our focus is reliability -- scaling the requests, while keeping stability. a lot of bug bash.

thank you for tuning into our journey. there's also a surprise later this week!!

[Eigen Labs @eigenlabs]

Read @sreeramkannan's recap on how anons beat Google in 72 hours

https://t.co/eDY8WP2afJ

[Eigen Labs @eigenlabs]

Our CTO @vishnu_patankar was on @OfflineOnAir recently, where he drew a fascinating parallel between AI and nuclear technology.

When a technology becomes powerful enough, the question is no longer just who builds it. It becomes who gets access to it. AI cannot remain concentrated in the hands of a few.

For it to become a true equalizer, access to intelligence has to spread.

[Gajesh @gajesh]

Wow. We closed in June 20th a little over 500M tokens served.

I guess, the next mark is 1B tokens in a day. We have found more efficiencies in the network where our providers can do 2x more capacity + concurrency.

The world's sleeping compute is waking up!!

[Gajesh @gajesh]

the beauty of building an open network is you get to interact and work with every type of unique individual

[Erick @ericktrify]

your Mac makes money while you sleep and you don’t even know it yet

[Soubhik Deb @soubhikdeb]

Everyone is worried about jobpocalypse but impact of AI on a given job depends on whether the job is low-dimensional or high-dimensional.

Companies are far more motivated to automate jobs with fewer tasks (low-dimensionality) because successful automation allows them to eliminate the entire labor sector for that role, resulting in greater cost savings.

In contrast, high-dimensional jobs, where only a few tasks can be automated, would still require human labor, making the investment case for total automation less attractive.

@alexolegimas has written detailed explanation of this dependency on the dimensionality of a job in his article https://t.co/vZJ6LbFulF.