Pinned Post
23 Dec 2025
419
945
7,561
5,274,778
Joined January 2022
- Posts 6,453
- Following 4,156
- Followers 33,944
- Likes 31,604
559 Photos and videos
Jaya Gupta reposted
Jun 6
We're introducing Simi, the fastest way to generate whiteboard-style explainer video with just a prompt or docs.
On the fastest tier, we can generate the video in 20 seconds.
500 Credits(~8 minutes) free for each QT & comment.
176
62
571
44,911
Jaya Gupta reposted
Jul 4
What a brilliant observation!
Big lab’ push for regulation may discourage their employees from leaving and starting new labs. You can not start easily in a regulated environments.
If that happens - there is no threat of them leaving and no need to pay top dollar for leading researchers.
Jul 3
The researchers getting rich off Anthropic secondaries are cheering for the thing that would make them ordinary employees again.
Right now they are paid like NBA free agents because they are the labs’ most visible moat. The frontier labs are struggling to hold a durable, ownable edge: models get copied, undercut, or matched by cheaper and open rivals within months. So the real advantage lives in a few hundred people who know how to push the frontier, and who can also leave, raise billion-dollar, double tranched seed rounds, and compete directly.
That is why the labs are paying them not to leave. with secondaries as retention payments, mission / fear, etc...
Pharma shows where this can end up. In a drug company, the value does not belong to the scientist. The scientist can be paid well, but not hundreds of millions over three or four years, because the durable value sits in the patent and the FDA approval. The researcher who discovered the molecule can quit tomorrow, but the company still owns the asset.
A regulatory moat would do something similar for AI labs. It would move value from the person to the institution.
Regulation is a wall against three threats at once: competitors, open source, and the labs’ own researchers. The researchers getting rich off secondaries today are, by cheering the regulated future, voting to end the exact leverage that made them rich.
2
23
5,720
Jaya Gupta reposted
Jul 3
Well said
Jul 3
The researchers getting rich off Anthropic secondaries are cheering for the thing that would make them ordinary employees again.
Right now they are paid like NBA free agents because they are the labs’ most visible moat. The frontier labs are struggling to hold a durable, ownable edge: models get copied, undercut, or matched by cheaper and open rivals within months. So the real advantage lives in a few hundred people who know how to push the frontier, and who can also leave, raise billion-dollar, double tranched seed rounds, and compete directly.
That is why the labs are paying them not to leave. with secondaries as retention payments, mission / fear, etc...
Pharma shows where this can end up. In a drug company, the value does not belong to the scientist. The scientist can be paid well, but not hundreds of millions over three or four years, because the durable value sits in the patent and the FDA approval. The researcher who discovered the molecule can quit tomorrow, but the company still owns the asset.
A regulatory moat would do something similar for AI labs. It would move value from the person to the institution.
Regulation is a wall against three threats at once: competitors, open source, and the labs’ own researchers. The researchers getting rich off secondaries today are, by cheering the regulated future, voting to end the exact leverage that made them rich.
1
5
2,935
This is a very good take
Jul 3
The researchers getting rich off Anthropic secondaries are cheering for the thing that would make them ordinary employees again.
Right now they are paid like NBA free agents because they are the labs’ most visible moat. The frontier labs are struggling to hold a durable, ownable edge: models get copied, undercut, or matched by cheaper and open rivals within months. So the real advantage lives in a few hundred people who know how to push the frontier, and who can also leave, raise billion-dollar, double tranched seed rounds, and compete directly.
That is why the labs are paying them not to leave. with secondaries as retention payments, mission / fear, etc...
Pharma shows where this can end up. In a drug company, the value does not belong to the scientist. The scientist can be paid well, but not hundreds of millions over three or four years, because the durable value sits in the patent and the FDA approval. The researcher who discovered the molecule can quit tomorrow, but the company still owns the asset.
A regulatory moat would do something similar for AI labs. It would move value from the person to the institution.
Regulation is a wall against three threats at once: competitors, open source, and the labs’ own researchers. The researchers getting rich off secondaries today are, by cheering the regulated future, voting to end the exact leverage that made them rich.
2
4
38
18,590
Jaya Gupta reposted
Jul 3
Smart take.
Jul 3
The researchers getting rich off Anthropic secondaries are cheering for the thing that would make them ordinary employees again.
Right now they are paid like NBA free agents because they are the labs’ most visible moat. The frontier labs are struggling to hold a durable, ownable edge: models get copied, undercut, or matched by cheaper and open rivals within months. So the real advantage lives in a few hundred people who know how to push the frontier, and who can also leave, raise billion-dollar, double tranched seed rounds, and compete directly.
That is why the labs are paying them not to leave. with secondaries as retention payments, mission / fear, etc...
Pharma shows where this can end up. In a drug company, the value does not belong to the scientist. The scientist can be paid well, but not hundreds of millions over three or four years, because the durable value sits in the patent and the FDA approval. The researcher who discovered the molecule can quit tomorrow, but the company still owns the asset.
A regulatory moat would do something similar for AI labs. It would move value from the person to the institution.
Regulation is a wall against three threats at once: competitors, open source, and the labs’ own researchers. The researchers getting rich off secondaries today are, by cheering the regulated future, voting to end the exact leverage that made them rich.
12
16
448
272,176
Jul 3
The researchers getting rich off Anthropic secondaries are cheering for the thing that would make them ordinary employees again.
Right now they are paid like NBA free agents because they are the labs’ most visible moat. The frontier labs are struggling to hold a durable, ownable edge: models get copied, undercut, or matched by cheaper and open rivals within months. So the real advantage lives in a few hundred people who know how to push the frontier, and who can also leave, raise billion-dollar, double tranched seed rounds, and compete directly.
That is why the labs are paying them not to leave. with secondaries as retention payments, mission / fear, etc...
Pharma shows where this can end up. In a drug company, the value does not belong to the scientist. The scientist can be paid well, but not hundreds of millions over three or four years, because the durable value sits in the patent and the FDA approval. The researcher who discovered the molecule can quit tomorrow, but the company still owns the asset.
A regulatory moat would do something similar for AI labs. It would move value from the person to the institution.
Regulation is a wall against three threats at once: competitors, open source, and the labs’ own researchers. The researchers getting rich off secondaries today are, by cheering the regulated future, voting to end the exact leverage that made them rich.
36
53
745
341,141
Jaya Gupta reposted
Jul 2
It's interesting that founders find raising at very high valuations makes the company much more attractive to potential employees. You'd think employees would want "cheap stock" (and more of it) that will become immensely valuable after they join. I guess the signaling value of a high valuation is often more powerful than the potential economics of "cheaper" equity.
49
7
224
64,010
Jaya Gupta reposted
Jul 2
You're wasting FLOPs when scaling inference compute: by independently sampling parallel attempts, you burn compute rediscovering the same solutions.
Introducing QuasiMoTTo: we scale parallel sampling with correlated samples instead! These samples have higher coverage, are marginally exact draws from the LLM, and can be generated in parallel.
Result: same performance with 25-47% fewer samples in test-time scaling 50% fewer training steps in RL!
In our new paper, we explore the design space of correlated samplers. Work with co-authors @probablynotaz9 (co-lead), @gandhikanishk, @noahdgoodman, and Emily Fox!
14
67
268
67,048
Jul 1
The best outcomes for some of the Neolabs will be to turn into quant trading firms
19
3
195
27,276
Jun 30
Seems topical again
3
8
122
24,631
Jun 29
:)
Jun 28
Best post on the topic of building an iconic company in the age of AIz
4
16
7,515
Jaya Gupta reposted
Jun 28
Best post on the topic of building an iconic company in the age of AIz
29
55
883
494,549
Jaya Gupta reposted
Jun 28
The advice here is timeless. And a great reminder that in this exciting moment where everything seems different, core principles still hold true
Jun 28
Best post on the topic of building an iconic company in the age of AIz
9
3
70
34,802
Jaya Gupta reposted
Jun 28
"The opportunity now is not to become the next OpenAI, Anthropic, Google, Palantir, or Tesla. But to ask what kind of company has not been possible before, and what kind of person has been waiting for it to exist."
~@JayaGup10
3
6
57
12,303
Jaya Gupta reposted
Jun 22
Trained some terminal agents with friends!
Introducing Tmax, open RL terminal agent models. Under default settings and shorter length (65k) token budgets, tmax outperforms prior open work on terminal use. We are releasing all data weights rollouts publically!
13
114
899
311,244