Rob Toews

Partner @RadicalVCFund, AI columnist @Forbes. "the machine does not isolate man from the great problems of nature but plunges him more deeply into them."

[Rob Toews @_RobToews]

10 (bold) predictions for AI in 2026:

1⃣ Anthropic will go public. OpenAI will not. 📈

2⃣ Details of SSI’s research and technology will leak to the public. The big labs will make meaningful adjustments to their research roadmaps as a result. 🤫

[Rob Toews @_RobToews]

preach

[TBPN @tbpn]

.@GavinSBaker explains how most people don't understand how hard it is to stand up energized compute, and that a lot of 'amateurs' will get cleaned out by 2027.

"Standing up a big GPU cluster is like running an F1 team."

"A bunch of people who have no relevant experience have acquired some land and power, then they get a zero down payment offtake agreement from someone, and a lot of that is being cleaned up."

"This is a forcing function to allocate GPUs to people who can actually bring on a cluster, energize the GPUs, and run it well."

"Companies like CoreWeave, Crusoe, Nebius, SpaceX, etc."

"It takes most people 3-6 months to get the GPUs energized. If we can, through Jensen's new DSX product, get that time down, I think that that's also healthy because that effectively takes a lot of working capital out of the market."

[Rob Toews @_RobToews]

@lightspeedvp Have you heard of @CrusoeAI?

[Andy Zeng @andyzengineer]

if you look closely at the reliability test later in the video,

you’ll see a few examples of “physical reasoning” -- it tries something, it doesn’t fit (because physics), so it backtracks, switches gears, then does something entirely different to reach the goal.

in this case, it’s taking out the stuck puck, making space elsewhere, and putting it back in.

things like this (… a sort of physical problem-solving with step-by-step “chain-of-actions”) should just work with models out of the box.

[Doris Tsao @doristsao]

1/ We discovered something totally unexpected about the neural structure of consciousness that overturns decades of thinking on its biological basis and brings us significantly closer to the recipe for generating conscious awareness.
New preprint: https://t.co/K8BwiHXvXg (led by @HesseJanis, with @FrankLanfranchi, @_conor_mcgrory, @Yuelin_Shi, and @EngelTatiana)

[Rob Toews @_RobToews]

Great discussion!

[Rebecca Bellan @RebeccaBellan]

One panel down at this year's @TechCrunch Disrupt! What happens when OpenAI ships your roadmap?

The audience was packed! And no wonder with so many founders looking for ways to defend their moat. Thanks to @YayLT @MichelTricot @_RobToews for your insights
https://t.co/9iG9ScJq30

[Rob Toews @_RobToews]

@mhdempsey precisely

[Rob Toews @_RobToews]

👀👀

[Generalist @GeneralistAI]

We're expanding access to GEN-1.5

Sign up to get in line
https://t.co/kaM93SJkXM

[Sarah Catanzaro @sarahcat21]

Nothing. Actually nothing. moves the needle on model performance as much as improvements in data quality. You probably can’t trust your vendors, your data is probably a mess. You need tools to curate data yourself. @datologyai is making this possible.

[Rob Toews @_RobToews]

🚀🚀🚀🚀

[DatologyAI @datologyai]

Today, we’re excited to launch the Datology Curation Studio, customizable frontier data curation for every team training its own models.

Data quality is the ultimate compute multiplier, and Curation Studio makes it easy for everyone.

[DatologyAI @datologyai]

Today, we’re excited to launch the Datology Curation Studio, customizable frontier data curation for every team training its own models.

Data quality is the ultimate compute multiplier, and Curation Studio makes it easy for everyone.

[Rob Toews @_RobToews]

ICYMI @datologyai is COOKING 👨‍🍳

[Maximilian Böther @MaxiBoether]

At @datologyai we run thousands of data experiments. Two things keep biting us:

- Same run on 16 vs 8 GPUs → different evals
- Every pipeline change → re-materialize the dataset

Introducing Zephon, the data loader we built to process data on the fly, deterministically. 🧵

[Maximilian Böther @MaxiBoether]

At @datologyai we run thousands of data experiments. Two things keep biting us:

- Same run on 16 vs 8 GPUs → different evals
- Every pipeline change → re-materialize the dataset

Introducing Zephon, the data loader we built to process data on the fly, deterministically. 🧵

[Will Manning @willmanning]

fast, distributed determinism is hard. really cool work from our friends at @datologyai!

(also... integrates with @vortexdotdev 👀)

[Generalist @GeneralistAI]

Robot puts a ring on it

GEN-1.5 puts an O-ring on a hydraulic rod gland, then takes the ring off, then puts the ring back on, then takes the ring off, then ring on, ring off, ring on...

Read more about GEN-1.5 in our blog post in the comments below.

[Dani Yogatama @DaniYogatama]

really proud to share what the team has been working on.

rho-1 is our 19B omni model. reasoning, real-time video simulation, and continuous robotic control inside a model.

we are very early on the compute curve and candid about current limits in the post, but the compounding gains across modalities are already clear.

[Harry Stebbings @HarryStebbings]

Why do you have to be vertically integrated to be a data center provider today?

“The supply chain to support large-scale AI data centers is sort of like a game of Whac-A-Mole. At different points, different bottlenecks come into play.

One of the key bottlenecks was a power distribution center. The lead time for this one component was 100 weeks, and I said, ‘Well, I do not have 100 weeks.’

We had consciously vertically integrated electrical manufacturing. We were able to do it in 28 weeks, and having the resources internally gives you a lot more flexibility to unblock these key bottlenecks.” @ChaseLochmiller

Love to hear your thoughts @stephenbalaban @MattLoszak @BrianGitt

[Jordan Jacobs @JordanJacobs10]

Congratulations @drfeifei and the @theworldlabs team! Very proud to have been on this journey with you since (before) the beginning.

[Rob Toews @_RobToews]

@cyantist @ChaseLochmiller @CrusoeAI winning

[Beff (e/acc) @beffjezos]

This company is going to rip to $100B easily.

Wish I could invest.

[Rob Toews @_RobToews]

Congrats to @CrusoeAI on its massively oversubscribed $3.9B Series F!

Crusoe will be one of the generationally defining companies of this AI era. We at @radicalvcfund are excited to double down in a big way.

https://t.co/ToAthfs1yG