Kimi K2: Hype, Benchmarks, and AI Trends
Key Points
- The episode opens with a round‑table of AI experts who debate whether the new open‑source model Kimi K2 is over‑hyped or under‑hyped, noting that while benchmark scores look impressive, its real‑world generalization remains unproven.
- Kimi K2, launched by the Alibaba‑backed startup Moonshot, claims to surpass Claude and GPT‑4 on coding benchmarks, sparking excitement that an open‑source model can now compete with industry giants in specialized tasks.
- The hosts caution that benchmark victories alone don’t guarantee broader utility, emphasizing the need to see how the model performs in diverse, production‑level scenarios.
- The conversation also touches on broader AI infrastructure trends, including Google’s massive new data‑center investment and Lawrence Livermore National Laboratory’s recent shift toward cloud‑based AI workloads.
- A brief retrospective on the “R1” project wraps up the episode, highlighting lessons learned from earlier AI initiatives and how they inform current developments.
Sections
- AI Podcast Discusses Kimi K2 - In the opening of the “Mixture of Experts” podcast, host Tim Hwang and guests preview AI topics—including a Google data‑center investment and Lawrence Livermore’s cloud adoption—before debating whether the Kimi K2 model is over‑ or under‑hyped.
- Cautious Optimism on New Model - Participants debate the hype versus actual performance of a new open‑source AI model, stressing the need for real‑world benchmarks and acknowledging it still falls short of top proprietary systems like Claude and GPT‑4.
- Open‑Source vs Proprietary LLM Cost Debate - Participants debate the performance hype of a new model, contrasting it with proprietary APIs and highlighting shifting cost dynamics from per‑token fees to predictable compute expenses for business use.
- Challenges for New Coding Models - The discussion highlights how the entrenched install base and vendor lock‑in of existing proprietary AI coding models create economic and strategic barriers that prevent even superior new models from gaining market traction.
- OpenAI Delays Open-Source Model - The speakers discuss OpenAI’s indefinite postponement of its much‑hyped open‑weight model, question why major U.S. AI firms are lagging behind Chinese open‑source efforts, and reference competing models such as DeepSeek, Kimi K2, and Mistral.
- Open‑Source AI Model Challenges - The speaker discusses the limited emergence of competitive open‑source AI models despite market interest, notes DeepSeek’s current lead, and probes whether developing open‑source models requires a distinct discipline compared to closed‑source approaches.
- Regulatory Barriers Shape AI Adoption - The speaker explains that compliance and tooling hurdles limit R1’s enterprise uptake in the West, while Chinese startups and academic researchers drive its use, amid geopolitical tensions and efforts to circumvent U.S. controls.
- Shift Toward Safe, Efficient AI - The speaker describes a strategic pivot from pure model capability promotion to emphasizing safety, governance, and efficiency—positioning Western AI firms as trustworthy alternatives and questioning whether U.S. companies can adapt to resource‑lean, open‑source model development.
- Patrick Mahomes Effect on AI - The speaker argues that DeepSeek‑V3's openness fuels creative AI progress, yet current models still fall short of the extraordinary, “Mahomes‑level” breakthroughs that would truly astonish users.
- Google’s $25B Energy Investment Overview - The speakers note a 17% NVIDIA stock drop yet persistent GPU demand for large‑scale inference, then shift to highlight Google's $25 billion investment in Pennsylvania hydropower and the PJM interconnect grid.
- Balancing Data Center Growth with Grid Limits - The speaker critiques Google's holistic energy approach, stressing that expanding data centers will strain the power grid, require diversified sources like nuclear and hydropower, and could raise electricity costs for consumers, particularly in underserved Midwestern regions.
- Google Drives Industrial-Scale Renewable Energy - The speaker highlights how soaring AI energy demand and long power‑connection delays are forcing big tech like Google to become major renewable energy buyers and investors, spurring an industrial‑scale race in clean power infrastructure and its downstream impacts.
- Environmental Concerns of AI Compute - Speaker questions the energy and water impacts of AI data centers and muses about alternative solutions such as space‑based computing.
- AI Inequality and Claude Expansion - The speakers caution that powerful AI models may widen economic divides while announcing Anthropic’s Claude being deployed across Lawrence Livermore National Laboratory to aid thousands of scientists with complex data analysis and hypothesis generation.
- From Tools to AI Collaborators - The speaker admires emerging AI for scientific discovery, debates whether it remains a supportive desk tool or evolves into a fully autonomous research partner, expresses concern over hallucinations, and envisions future AI agents acting as co‑authors that generate hypotheses.
- Behind the Research Narrative - The hosts discuss overlooked aspects of scientific research, acknowledge the contributors, and close the episode with thanks and a podcast promotion.
Full Transcript
# Kimi K2: Hype, Benchmarks, and AI Trends **Source:** [https://www.youtube.com/watch?v=hNvbeXus-pM](https://www.youtube.com/watch?v=hNvbeXus-pM) **Duration:** 00:47:40 ## Summary - The episode opens with a round‑table of AI experts who debate whether the new open‑source model Kimi K2 is over‑hyped or under‑hyped, noting that while benchmark scores look impressive, its real‑world generalization remains unproven. - Kimi K2, launched by the Alibaba‑backed startup Moonshot, claims to surpass Claude and GPT‑4 on coding benchmarks, sparking excitement that an open‑source model can now compete with industry giants in specialized tasks. - The hosts caution that benchmark victories alone don’t guarantee broader utility, emphasizing the need to see how the model performs in diverse, production‑level scenarios. - The conversation also touches on broader AI infrastructure trends, including Google’s massive new data‑center investment and Lawrence Livermore National Laboratory’s recent shift toward cloud‑based AI workloads. - A brief retrospective on the “R1” project wraps up the episode, highlighting lessons learned from earlier AI initiatives and how they inform current developments. ## Sections - [00:00:00](https://www.youtube.com/watch?v=hNvbeXus-pM&t=0s) **AI Podcast Discusses Kimi K2** - In the opening of the “Mixture of Experts” podcast, host Tim Hwang and guests preview AI topics—including a Google data‑center investment and Lawrence Livermore’s cloud adoption—before debating whether the Kimi K2 model is over‑ or under‑hyped. - [00:03:17](https://www.youtube.com/watch?v=hNvbeXus-pM&t=197s) **Cautious Optimism on New Model** - Participants debate the hype versus actual performance of a new open‑source AI model, stressing the need for real‑world benchmarks and acknowledging it still falls short of top proprietary systems like Claude and GPT‑4. - [00:06:23](https://www.youtube.com/watch?v=hNvbeXus-pM&t=383s) **Open‑Source vs Proprietary LLM Cost Debate** - Participants debate the performance hype of a new model, contrasting it with proprietary APIs and highlighting shifting cost dynamics from per‑token fees to predictable compute expenses for business use. - [00:09:26](https://www.youtube.com/watch?v=hNvbeXus-pM&t=566s) **Challenges for New Coding Models** - The discussion highlights how the entrenched install base and vendor lock‑in of existing proprietary AI coding models create economic and strategic barriers that prevent even superior new models from gaining market traction. - [00:12:36](https://www.youtube.com/watch?v=hNvbeXus-pM&t=756s) **OpenAI Delays Open-Source Model** - The speakers discuss OpenAI’s indefinite postponement of its much‑hyped open‑weight model, question why major U.S. AI firms are lagging behind Chinese open‑source efforts, and reference competing models such as DeepSeek, Kimi K2, and Mistral. - [00:15:38](https://www.youtube.com/watch?v=hNvbeXus-pM&t=938s) **Open‑Source AI Model Challenges** - The speaker discusses the limited emergence of competitive open‑source AI models despite market interest, notes DeepSeek’s current lead, and probes whether developing open‑source models requires a distinct discipline compared to closed‑source approaches. - [00:18:43](https://www.youtube.com/watch?v=hNvbeXus-pM&t=1123s) **Regulatory Barriers Shape AI Adoption** - The speaker explains that compliance and tooling hurdles limit R1’s enterprise uptake in the West, while Chinese startups and academic researchers drive its use, amid geopolitical tensions and efforts to circumvent U.S. controls. - [00:21:52](https://www.youtube.com/watch?v=hNvbeXus-pM&t=1312s) **Shift Toward Safe, Efficient AI** - The speaker describes a strategic pivot from pure model capability promotion to emphasizing safety, governance, and efficiency—positioning Western AI firms as trustworthy alternatives and questioning whether U.S. companies can adapt to resource‑lean, open‑source model development. - [00:25:10](https://www.youtube.com/watch?v=hNvbeXus-pM&t=1510s) **Patrick Mahomes Effect on AI** - The speaker argues that DeepSeek‑V3's openness fuels creative AI progress, yet current models still fall short of the extraordinary, “Mahomes‑level” breakthroughs that would truly astonish users. - [00:28:18](https://www.youtube.com/watch?v=hNvbeXus-pM&t=1698s) **Google’s $25B Energy Investment Overview** - The speakers note a 17% NVIDIA stock drop yet persistent GPU demand for large‑scale inference, then shift to highlight Google's $25 billion investment in Pennsylvania hydropower and the PJM interconnect grid. - [00:31:28](https://www.youtube.com/watch?v=hNvbeXus-pM&t=1888s) **Balancing Data Center Growth with Grid Limits** - The speaker critiques Google's holistic energy approach, stressing that expanding data centers will strain the power grid, require diversified sources like nuclear and hydropower, and could raise electricity costs for consumers, particularly in underserved Midwestern regions. - [00:34:39](https://www.youtube.com/watch?v=hNvbeXus-pM&t=2079s) **Google Drives Industrial-Scale Renewable Energy** - The speaker highlights how soaring AI energy demand and long power‑connection delays are forcing big tech like Google to become major renewable energy buyers and investors, spurring an industrial‑scale race in clean power infrastructure and its downstream impacts. - [00:37:48](https://www.youtube.com/watch?v=hNvbeXus-pM&t=2268s) **Environmental Concerns of AI Compute** - Speaker questions the energy and water impacts of AI data centers and muses about alternative solutions such as space‑based computing. - [00:40:51](https://www.youtube.com/watch?v=hNvbeXus-pM&t=2451s) **AI Inequality and Claude Expansion** - The speakers caution that powerful AI models may widen economic divides while announcing Anthropic’s Claude being deployed across Lawrence Livermore National Laboratory to aid thousands of scientists with complex data analysis and hypothesis generation. - [00:43:58](https://www.youtube.com/watch?v=hNvbeXus-pM&t=2638s) **From Tools to AI Collaborators** - The speaker admires emerging AI for scientific discovery, debates whether it remains a supportive desk tool or evolves into a fully autonomous research partner, expresses concern over hallucinations, and envisions future AI agents acting as co‑authors that generate hypotheses. - [00:47:08](https://www.youtube.com/watch?v=hNvbeXus-pM&t=2828s) **Behind the Research Narrative** - The hosts discuss overlooked aspects of scientific research, acknowledge the contributors, and close the episode with thanks and a podcast promotion. ## Full Transcript
It's all great in theory, but
then you know what happens when it comes to.
I've got a power Google AI overviews.
Or are Mr. and Mrs. Jones down the road?
Need to watch the television this evening or need to keep warm in the winter?
And you're like, uh, I'm paying for the data center.
Sorry, grandma. We have a pre-training run.
All that and more on today's Mixture of Experts.
I'm Tim Hwang, and welcome to Mixture of Experts.
Each week, MoE brings together
a crack team of the most brilliant and entertaining
researchers, product leaders, and more to distill down and chart a path
through the ever more complex landscape of artificial intelligence. Today,
I'm joined by Abraham Daniels, Senior Technical
Product Manager for Granite.
Kaoutar El Maghraoui, Principal Research Scientist and Manager for Hybrid
AI Cloud, and Chris Hay, Distinguished Engineer.
We have a packed episode today we're going to talk about
a little bit of a retrospective for R1.
We'll talk about a huge data center investment by Google.
We'll talk about the adoption
of cloud by Lawrence Livermore National Laboratory.
But today I actually want to start first with Kimi K2.
And I think for our round the horn question,
we'll do a really simple one, which is Kimi K2.
Is it over hyped or under hyped? Uh,
Abraham, curious. Have you got any thoughts on that?
Honestly, I don't know. It's,
uh, from a benchmark perspective,
it looks amazing, but I think we have to wait and see.
From a generalization perspective, it's actually as good as they say.
All right, Chris, what do you think?
It is actually really good.
But it's not better than Claude.
No matter what the benchmarks say.
All right.
And finally, last but not least, Kaoutar. What do you think? Yeah,
I think it's a little overhyped,
but yes, it's a it's a very good model. Okay.
A lot to get in here too.
I love these opinions. They're like ah,
maybe good, maybe bad. Um,
so just give a quick background
for folks who may have not been watching this.
So, Kimi K2 is a new model
that dropped from the Alibaba backed startup moonshot.
And it's an open source model notably.
And it's been kind of really storming the charts.
There's been a lot of chatter about it online.
People are saying it's the best thing since sliced bread.
And I think the most interesting thing about the launch
is that the Moonshot
company has basically claimed
that against benchmarks, it is surpassing
the latest state of the art for Claude and GPT-4,
particularly on coding benchmarks, which is a big deal, right?
The idea that on this specialist task of coding this open
source model is now challenging,
you know, the biggest players in the game.
And Abraham, maybe I'll start with you
because I thought your response was maybe a good way into this discussion.
You were saying, well, hey, it looks great,
but we actually don't know yet if it's any better.
Um, what do you mean by that?
Tell us. Tell us more. Well,
a couple of things. One, in public benchmarks, uh,
you know, as we've spoken, a couple in a number of these, uh,
you know, Mixture of Experts, episodes can be gained
and they don't always tell the full story. Um,
so although, you know,
they may have published that they're better than Claude and GPT
until we can actually get some independent
or third party or see what the community actually thinks.
think it's, you know, maybe the claim
is a little bigger than it really is.
Um, also, it's,
uh, I don't know my opinion
that there's a lot of kind of craze at the beginning,
and then things kind of settle down
and we kind of figure out where it really stands.
So I'm cautiously optimistic about its performance,
but I'd like to just see some, uh, real world applications,
whether that's, you know, integrating to certain stacks or, you know,
actually demonstrating side by side comparisons,
whether this is actually as good as they say it is. Yeah. For sure.
And Chris, I think maybe I'll turn to you next.
Like, you know, I think the caution is well warranted.
And I think at this point I like barely look at like the benchmarks
in the blog post when they announce models because I'm like,
ah, it's all a gamble. It's all a trash.
But you seem to be convinced you're like like just on playing around with it
It's a good model, but it is definitely not
as good as Claude and GPT-4. What,
what leads you to say that, um,
putting my hands on the keyboard and typing stuff in and see what comes out?
Give me more than that, though. Of course.
But this is more than just a vibe check, right?
You actually think against certain tasks.
You think that, like, still,
this is not surpassing the state of the art here.
No, I don't think. I don't think so.
So the first thing I would say it is by far,
in my humble opinion,
it is the best open source model out there at the moment.
Or open model.
Um, it they have done a phenomenal job.
I mean, it's a 1 trillion parameter model.
So this thing is big, okay.
It is a mixture of expert model with a lot of models,
but it's still a big model.
And you know, and you need a lot of disk space
to get that running on your machine.
Now.
Um, but it is the best model for an open source,
but it doesn't be closed.
And there are a lot of things
that I think are really good for this model.
I mean, when I was playing with it, I really liked its planning capability.
I really liked its tool use.
So this is a model that is definitely being designed
for a genetic behavior, right?
They've really focused on the plan and they've really focused on the use of tools.
Um, and I think that is going to be exciting
when we run a smaller model
because to be honest, when you want to run agents,
I said agents, of course, but when you want to run agents,
you want your models to be small and fast and lean
and and I think it's going to do a phenomenal job as well.
The other thing is it's not a reasoning model.
So it doesn't have that thinking capability yet.
They've just provided a base model and a model. Um,
but it is fabulous for the chat.
So code wise or to sort of
come back to what I said, there are ten, right?
Code wise I think open source, open weight model,
It is the best coding model out there.
I've used pretty much every single one of these models,
whether it's the Qwen models, whether it's a DeepSeek, etc.
it really is the the best coding model out there for an open model,
but it doesn't be closed, right?
It may be that on the benchmarks.
And back to Abraham's point, right.
is a lot of these things are gained
towards the benchmarks to try and get that sort of edge.
But when you put it in real coding scenarios, right.
Um, I want to code up this,
this particular program, change this, do this or whatever.
It does a good job.
But the code is better, right?
I mean, cloud is giving me better results
than I'm seeing from my vibe checks.
Um, but but fair play to them.
I don't take anything against that
It is an incredible model.
And for kind of the budget, the compute,
the time that they've had again, spectacular. Kaoutar,
I know you came in basically saying that you felt like
it was a little bit of an over hyped launch. Um,
and so do you kind of buy Chris?
You're basically like very good, but still like, you know,
as compared to the proprietors.
You know, I think it's still still lagging a little bit behind.
Yeah. So but I think, you know,
um, there are also other angles that, you know, this,
um, release or this launch is kind of,
uh, getting us to start thinking about
which is more on this evolving war on the cost,
you know, the open source versus, you know, the proprietary APIs.
So if you look at companies like OpenAI
or, you know, Anthropic or Google,
they're charging per token for API access. But,
you know, these open source models with models
like Kimi K2, Llama or Mistral,
you know, the cost here is shifting from these API
free to a fixed or at least predictable infrastructure cost,
you know, so you're paying more for the compute.
it's like we're getting, you know,
with models like, you know, it is a great model.
I played with it a little bit, uh, but it's kind of where
the good enough tipping points.
So for, you know, many business tasks,
you know, you know, summarization, classification and etc.
these open models are doing a pretty good job,
you know, even superior than, you know, the closed ones.
So now I think we're kind of getting into this phase
where companies, you know, can now adopt,
you know, this hybrid strategy, use maybe expensive
proprietary models for maybe complex frontier tasks,
but then offload the bulk of their workload
to really cheaper or self-hosted open source models.
So but I also feel, you know, with, you know, this launch,
we're kind of getting into this maturation of the open source AI movement.
I mean, it just didn't happen with Kimi K2,
but also with the other, you know, open source models.
So it's no longer about, you know, providing a free alternative,
but also about competing directly on
performance and features with these,
you know, other, you know, closed source models.
So I think with this release, it's also kind of pushing for, you know,
kind of, um, towards putting more pressure on the pricing models
of these proprietary giants like OpenAI and Google
and, you know, kind of, you know, putting a lot of pressure on them.
So the the future of these, you know, enterprise
AI is not just a single vendor solution.
I think it's kind of leaning toward more cups,
optimized portfolios, hybrid models. And,
you know, I think Kimi K2's success
or, you know, really great performance signals that, you know,
the primary battleground in AI here is shifting
from this pure performance race to kind of a war of economic efficiency.
And also like the strategic control. Yeah.
And I think I did want to pick up on the strategic control point. Um,
there's an interesting observation that some people are making, which is, okay,
I know this group is maybe a little skeptical about sort of K2's
ultimate capabilities on coding,
but like, assume for a moment that it is actually better
than what Claude and like, say, OpenAI can provide.
A lot of people were pointing out that actually now it's actually
a little bit difficult for Kimi to compete in that universe,
because a lot of people are on platforms and endpoints
that are using all the existing
leading proprietary models.
And I guess, Abraham, maybe I'll throw it to you
because I know you're working with Granite day in and day out.
Like, do you think that there's kind of this really interesting dynamic emerging where
now the kind of like preexisting install base effectively,
if you will, for these models, particularly in coding
like means that it's actually really difficult for like a new model,
even if it's better to get in and actually compete with these proprietors.
Do you buy that at all? I'm not really.
I think it's less about whether if it's better or not.
And to to Kaoutar's point, it's really
what are the economics of using this model versus,
you know, vendor lock in or locking into a particular stack or infrastructure.
I think the question really, is it good enough
where the price tag aligns with our,
you know, our business case or, you know, our user base?
And I think you're consistently seeing that these, you know,
open source is now a strategic weapon as opposed to just a, um,
you know, a mandate by an organization
where you're starting to disrupt a lot of these, you know,
closed source models, and when you can actually brush up against their performance, whether that's,
you know, R1 on reasoning or Kimi K2
or communicate to on on coding,
you're really signaling to the market that, you know, um,
one vendor lock, in
my opinion, of Interlochen, was always going
to be kind of dismantled as you had a proliferation of developers.
was going to kind of erase to the bottom,
but this really just kind of expedites it.
Uh, and then too, I think there's also, you know,
developer centric pricing is going to continue to force a downward pressure.
I think over the last six months, you've seen an actual explosion
in cost per input and output tokens.
So I be personally, I think this is amazing. Granite,
as a, you know, as a model,
we are huge proponents of open source
licensing with, you know, no, not all things open sourcing.
So I think this is the right direction, not only for the field. Um,
and then I also think this is kind of signaling
to, Llama and OpenAI that they have to start
to take this very seriously in terms of how this takes into their roadmap, too.
So with OpenAI starting to
or hinting at another open source model, the first since GPT-2.
So firstly, um, you know, back to your question,
I yeah, I don't think this is necessarily an issue.
I think this is really just a it's an economics question, more
so than, uh, a technology question.
The second topic of today that I really wanted to get into
was zooming out from Kimi K2, right.
Someone pointed out to me recently where
six months since the R1 launch,
and which is amazing because R1 launched January 20th, 2025,
it already feels like it was six years ago, not just six months ago. Um,
but I think it might be good for us to kind of
just talk for a few minutes, zooming back a little bit on
like what has changed since R1 launched.
Um, and I think Abraham, you're picking up,
I think, on one thing that I did want to bring up,
which is, you know, in the midst
of all this, OpenAI announced
that it would be kind of delaying indefinitely
the launch of its open source model, which was kind of way
hyped and was originally read as kind of a response
to this new generation of Chinese open source models,
but now appears to be kind of like on the on the back burner
Well, back burner is maybe the wrong word,
but delayed for an unknown amount of time. Um,
I guess, Chris, maybe to throw it to you like,
you know, do you feel like the US companies in
some ways have, like, not been able to kind of like,
answer this open source challenge at all? Right. Like,
I think in some ways like Meta is still competing,
but OpenAI is not really open sourcing.
It feels like there hasn't been another kind of marquee model that says, okay, actually,
a lot of these kind of dominant US companies
can kind of keep up in this race.
I think there's different
economics and power shifts in play in this sense.
I don't think there's any reason why OpenAI or Anthropic
can't release an open weight model. Right?
Um, they're obviously choosing to do other things there.
I stick by my statement that I said earlier for size.
I think the best open weight
models out there are the deep six.
That is the um, you know, now surprised by became,
you know, suppressed by the Kimi K2 model.
Um, you know, the Mistral models are incredible.
Their open weight models, um, you know,
especially their 24 billion parameter one and the Mistral medium,
they they're really great models. Um,
and I love what we're doing with Granite
with the 7B models or in the 8B models. Right.
Everybody. And the, the the 1B models,
I think everybody's forgetting about these really small models..
And and actually they become super important especially for things like agents.
So I, I think they're missing a trick.
I mean, the only American company
that's really producing good open weight
models is Google it that, you know,
Google at the moment and, and IBM obviously.
But I mean, on the kind of the higher number of parameters.
Um, so I just think
I think there is more to do in that effort.
And and that's the you know,
it's running away from there.
So I'd like to see that position change.
Um, because the reality is
there's a risk for all these companies,
which is once you start to get, um,
competitive models
and you're not going to compete with a trillion parameter model
But if you can get a really great coding model
down to the 8 billion parameter number and and and again,
I don't think that's far off when you think about some of the like,
Mistal's doing with the 24 billion parameters
then are about to counter his point about cost economics.
If I can run something on my laptop
and I can get good code from it, or I can run
good agents from it, that starts to affect their business model.
So I, I'm a big fan of open weight, a big fan of open source.
I really like to see all the close, uh,
source providers open up their models and open up their weights.
I'd like to see that as, uh, just get it done. Yeah.
For sure. Well,
and I think that's that's one thing
I did want to get to in and counter.
I guess you've been name checked, so I'll kind of bring the conversation
back to you is like, you know,
there's obviously different economics and the kind of sort of us
leading companies are like trying a couple different things in the space.
Um, but it is kind of interesting to me
that it feels like the number of kind of like,
like, I guess when I think about open source, I think like,
oh, well, there's going to be tons and tons of different players
putting out lots and lots of different models.
And, you know, we're going to see this space really, really kind of open up.
I mean, to Chris's point, you know, even though the Chinese market has kind of
like really invested in open source, it still feels like after six months,
DeepSeek is really still kind of like in the lead here. Right.
Like that actually, we haven't seen like an explosion of new companies
offering open source models in the space that are kind of
at least as competitive.
I guess the question I kind of want to get you to respond to
is like whether or not you think there's like a special discipline
with doing open source models that's maybe different from closed source.
Like, is there like a different style of what's going on here
that actually is almost as difficult as doing a closed sourced model
well. Yeah, that's a very good question, I think. Um,
what really, you know,
I think help seek is the efficiency
or, you know, aspect of it.
So I think the key innovation was mostly behind
their architectural efficiency
where they employed, you know, the, you know, the bag of techniques
of mixture of experts, reinforcement learning,
you know, optimizations all the way to the level of the PTA, etc..
So that was I didn't think that was kind of a
maybe, uh, a kind of a breakthrough thing, but more,
you know, efficient implementations
and clever ways of using existing techniques.
So, and, uh,
and, of course, you know, there is an ongoing debate
about the nature of, you know, Deep Six achievement.
You know, while, you know, some of them view that, you know,
their methods are revolutionary breakthroughs.
You know, I'm more, you know, along the lines of those that you know,
that think that, you know, it's a clever and effective implementations of existing techniques
rather than a fundamental paradigm shift. So.
So but that was I think the efficiency was very important because, you know,
showing that you can get, you know,
to these state of the art models with less costs.
That was I think, a very important shift.
You know, that they have showcased here.
Um, and since then we've seen,
you know, many releases where they kept improving their models.
So they have the steady, you know, flow of releases where they kept improving.
So that that was really great to see.
Um, so going to your question,
what's kind of the maybe the the recipe here?
I think, uh, of course, you know,
being able to, to be state of the art
kind of beating these benchmarks,
but also having the capability to do these things efficiently.
Um, but if you see like six months,
you know, from their launch,
um, have they kind of shaken the markets?
Uh, have they kind of, you know,
uh, like, especially the closed source ones.
Uh, probably not that much.
You know, the enterprise uptake for,
you know, deep sky field, it still remains a limited.
And I think it's mostly due to regulatory and compliance
and some of the tooling blockers.
So the adoption, of course,
uh, is mostly concentrated in Chinese
based startups and hobbyist communities.
But then, you know, if you look at, you know,
in the, the West and the, in the US and, you know, so the, the,
the enterprise uptake is still limited.
Um, and it's mostly, I think,
of course, in the academic and the specialized domain,
there is a lot of traction here.
Like a lot of researchers are leveraging R1
for math, problem solving, code generation
And, you know, especially for the Chinese language, medical diagnostics, for example.
But then, you know, in the enterprise,
I think it's I feel it's still limited. And,
uh, maybe that's also kind of part
of this geopolitical, geopolitical AI
race where we've seen it is getting intensified.
So because deep seeks open source
strategies, encouraging, you know, the rivals for example,
moonshot are like we're seeing with Kimi K2 to follow here
and uh, and especially to kind of trying to partially bypass
the US controls or the US chip controls,
so that that is really something that is so important for them.
Um, but you know, what we see also on the Western governments,
they're really trying to double down on these trustworthy AI frameworks,
uh, which is becoming very important.
Yeah. I think this is so a lot to unpack there.
And I think you're getting to something I think is really interesting is
I think the narrative when R1 launched was,
oh, man, all of these American companies are suddenly in trouble
because you have this incredibly powerful model
and it's available for for free, right.
And I think six months on, my kind of reflection is caps
are the same as yours, which is actually like enterprise
adoption has been kind of less than I would have thought.
Um, and and that's, that's pretty interesting, right?
That like, in some ways, the market dynamics
that we kind of originally thought with R1
and particularly around open source
don't necessarily seem to be playing out the way we thought.
I guess, Abraham, do you have any responses to that?
Like, it's it's kind of odd to me
that, like, you have this incredibly great model
that's like available for free
and we just haven't seen like mass adoption in a six month period.
Like, if anything, you know, the proprietary,
like your open eyes or anthropic of the world seem,
you know, they're changing the strategy, but they're not like
completely demolished as a result of this change.
Yeah. And I think that's exactly it.
I think it was less of like, uh, like competition with respect to,
you know, another model that's in the queue
in terms of, you know, what your enterprise is going to use.
I think it was just more of like a what the strategy was.
The status quo pre R1 shifted to be able
to differentiate from R1.
So you know where those models were clearly ahead,
you know open weight was able to give you
parity on key resource key reasoning tasks.
So it shifted to you know let's get smarter
cheaper inference as the goal. Uh,
you know, agents were agent tech orchestration was already
kind of, you know, bubbling up, but everybody doubled down on,
you know, being able to develop
an LLM that was, you know,
a key supporter of agent tech workflows.
Safety was also doubled down to in terms of,
you know, our models are, you know, safe from a,
you know, red teaming from a governance from an AI perspective
both on the model and the data side.
So I think it was really just a,
um, a shift in strategy from like a,
uh, a model capability,
PR, if you will, perspective in order
to differentiate from our one,
to showcase that, you know, we are moving forward as a,
you know, US based or Western model developer companies,
um, and less of a,
you know, R1 was now considered a viable option
as part of like an enterprise use case.
That's really interesting. Chris,
maybe a final comment. Again,
pulling out of kind of Qatars theme.
You know, I think, Kaoutar, you pointed out, I think it's a really interesting thing,
which is, well, maybe part of Our one's genius
is its dedication to efficiency, right.
They were able to assemble all these hacks together to really get like, squeeze
a lot of results without having a whole lot of resources.
Um, and I think a little bit about, like, what it means
to be efficiency minded
and how it can be really hard
to kind of like, think in that style
if you're used to having like, the most compute
and the most money in the entire world.
Um, and I guess, Chris, I don't know
if there's almost kind of a thesis here
that I want to run by you, which is like,
could it be hard for American companies to pivot into this?
Which is a big deal if you think that, like, small open
source models are going to be like the future of agents.
Is it hard for these companies to pivot into this kind of efficiency mindset?
Because in some ways, technically, I think they're like
maybe so used to an environment where it's like,
we never have to think about
how to assemble all these things to squeeze
the most results out of limited resources.
Um, I'm curious about if you think that's almost like a barrier
in some ways to these companies pivoting towards open source.
I think that when you are limited by your resources,
you become super creative.
And actually, if we think about the Kimi K2 scenario,
they got super creative, right?
One of the biggest things that they did is
they came up with their, um, their
new optimizer. Right? The,
the Muon optimizer, which,
which was really about them being able to, um,
train very, very large models in a consistent way, um,
and not, uh,
basically have their training losses mess up during that process.
That is a huge moment. Now,
we don't know all the details behind that,
but the innovation there is great.
They've moved away from the optimizers others are using, right?
When I think about the deep seek moment
and their efficiency, they similarly.
But nobody really cared about deep seek.
But when they first launched anyway, DeepSeek-V3 came out in December.
But it wasn't until they released our one where we got excited,
and it's because they had, um,
the reasoning model, and it was pretty much close
to the old series of models there. Right?
And then they were open about how they published it.
They went through, um,
you know, their RL flow
and how they train the, the grp stuff, etc. and,
and we all learned stuff and it was all great, but they were innovative and,
and the great thing is they were open about it
and everybody's been running around copying their techniques and learning from them.
Kimi K2 wouldn't exist if DeepSeek-V3 wasn't open
about how they trained the V3 model,
so I think that in itself is going to boost that creativity.
But to your point, I'm not quite sure
if you're just sitting there with,
you know, hundreds or thousands of each one. Hundreds.
I'm not sure you're
you've got all the compute you need, right?
I'm not sure you're going to be.
So, you know,
you're just going to get your job done as opposed to like going,
oh, I can't do this because I don't have this
and I need to figure my way out of it.
So I think, I think that is helping them.
But but why is deep seek
maybe, you know, six months on to your point.
I'm going to call it the
I'm going to call it the Patrick Mahomes effect. Right.
There are great quarterbacks
kicking around Tom Brady who is the greatest.
And then great quarterbacks who come along and you go oh there's their golf.
They're like you know you're like oh okay.
Even Justin Herbert people who shoot me for that they'll go ah okay.
Do you know what I mean? Because you're not seeing anything amazing.
Over time you get used to them.
But then when you look at Patrick Mahomes play
and you're like, how did he do that?
No human on earth is able to make that through.
How did he do? He.
He wasn't even looking.
And and I don't think those models are quite doing that yet.
Right. Because the models that have come out
are equivalent or they're,
they're about the same
as the, the, the models
and nobody really cares about the same. Right.
If you think of a like,
you know, if you think of a Super Bowl or what,
nobody remembers who lost the Super Bowl.
They're close enough to the the team that won.
Right. But but people care about the winners the greatest. Right.
So I think for one of these to take hold
and really upset OpenAI and throw, pick etc.,
they're going to have to do something like the no model has ever done before.
It's just like, oh, I press a button and it's it's
created an entire billion dollar company overnight.
Wow. And it's done it on a chip that runs on my laptop.
That will be like, whoa.
I mean, they're not impressive.
I would be impressed.
Nobody's going to care at that point.
You think you're going to stick on?
I don't know, I'm going to stick typing in ChatGPT.
You're like, no, I'm running over the new thing.
I've got to see that.
Whereas if it's just like, ah, it's the same as it was before.
You're like, well, it's just the same.
I'll stick with what I've got.
That's what needs to change.
And also I think the, the first movers
adventure's always has a big, you know, effect.
You know, if you know, I think OpenAI, which, uh ChatGPT,
you know, kind of gained a lot of, you know, mass,
uh, adoption.
And so once you get used to that, you know,
sometimes switching from that environment to something else,
you know, you know, you really need to have like, Chris
say something completely kind of a wow effect,
something not just incremental.
And, and I think I'm going to back to your, uh, like the, the
resources or the compute question.
So even, you know, you know, our one kind of shakes,
you know, the, the GPU dominance, the NVIDIA GPU.
So the stock dipped, you know, significantly, like 17%.
But then the demand for NVIDIA hardware
kind of rebounded because large scale inference still relies a lot on GPUs.
So so I mean, we had the panic moments,
but then but the efficiency gains,
you know, really haven't negated the massive compute needs
that are still, um, still there.
Yeah, I think it's right.
Well, we'll be checking in again another six months, I think.
Kind of like using R1 as a peg and kind of moving out
I think is really useful just because the space moves
so, so quickly.
I can move us on to our next topic.
Um, announcement coming out of Pittsburgh.
Really big event this week.
Uh, the president was there.
All the major companies were there.
Um, but I think there's one announcement in particular I want to zoom in on,
which is, uh, Google announced
that it'd be making a $25 billion worth of be,
um, announcement to invest in energy infrastructure.
Uh, so for one part
hydropower in Pennsylvania,
and then also something that's known as the PJM interconnect, right,
which is a network grid that stretches across
New Jersey, Pennsylvania, West Virginia, Virginia,
really large area of the country. Um,
and, you know,
I think this is in some ways like taking a step back,
like both wild both in terms of the dollar amount being committed,
but also just to remind ourselves
that, like, Google is like a company that started doing search. Right.
And so it's not intuitively obvious that you would eventually say years later,
you know, we're going to be investing billions of dollars
in going all the way upstream to really, literally change,
like the energy grid of a whole part of the country.
Um, and so I guess
Abraham question for you is just like,
how far do you think this all goes? Right.
Like at some point, does Google just say we're going to be owning
and operating a nuclear power plant?
Like, it feels like in some ways, AI is generating such demand on the grid.
These companies really need to assure energy access.
And at some point, it kind of feels like, okay,
where this all goes is like vertical integration.
You can you can subscribe to have your energy bill sent to you from Google.
Is that where this is all going?
I mean, uh, it's a great question.
Um, I think you can
Microsoft and Meta have both,
you know, committed massive amounts of money to build their own data centers.
I think Google's taking a different approach in terms of not only building data center,
but what I think was missing with the
the prior ones is investing in the actual grid themselves,
as well as investing in the community around them.
Um, so to your question,
you know, maybe it kind of makes sense
if you talk about the actual cost of power
to be able to manage these data centers. Um,
if anything, I kind of, you know, clap to Google to
to actually take more of a holistic approach
in terms of being able to create,
you know, um, compute or to create data centers,
because I feel one thing that's typically missing is,
you know, getting a better understanding of how what the impact of these data centers
are to the surrounding, whether it's the grid, the,
you know, ecosystem, the
you know, this takes a ton of water to be able to cool these things.
the runoff, um, so I think from Google's perspective,
they did a more of a holistic approach, which I kind of applaud to you.
I think this is only can we continue to happen.
And you kind of mentioned, you know, nuclear energy.
Like I think the next step is really better understanding,
you know, whether it's hydropower, you know, electricity, nuclear is like,
where's all this energy actually going to come from?
Because, you know, depending on what you read,
you know, by 2030, data
centers are going to represent 1% to 3% of all power on the grid.
And right now it just can't, you know, support
that, let alone manage it.
So it's it's really kind of focusing on how do we support today.
And then how are these hyperscalers going to invest in the grid
if they're going to be the primary user of the energy coming off of it?
Um, because there are some downstream impacts.
And I mentioned, you know, an environmental.
But when you have all these data centers or these players, you know,
integrating to the grid that drives electricity costs up for, you know, your everyday consumer
and, you know, some of these areas that these grids are built by.
These data centers are built in Middle America.
Um, you know, these aren't areas
where, you know, you typically have,
uh, you know, access to as much as you would
maybe like in New York or Boston or in San Francisco.
So, uh, I think it's just important
to kind of take a little bit more of like a long tail view in terms of,
you know,
building out the grid
and building out these data centers and really focusing on,
you know, what are the impacts above and beyond,
you know, the business side of things and where the impacts from the,
you know, the surrounding community and environment.
Yeah.
Are you think about hardware a lot.
Um, and I think that, um,
you know, one of the things I love about AI is how it just kind of
inverts our sense of what's abundant and what's scarce,
you know, like, I think a few years ago
you would have said, oh, there's just so much data.
We're never going to run out of data.
And I think in AI land, we routinely have conversations where we're like,
how do we get the next most valuable tokens?
And it feels like for a long time,
at least in what we're talking about here,
like hardware felt like the real bottleneck,
which is like, can you get access to Jensen's chips?
That really was the big thing.
Over the longer run, though,
the midterm, like let's say 5 to 10 years, like,
do you think energy becomes the new bottleneck?
Like at some point I think like there will be more chips,
there will be more GPUs, there'll be more suppliers of those GPUs.
There'll be changes in models that maybe make the specific hardware less necessary.
But it kind of feels like maybe where this is going
is that whatever hardware platform
you use, the energy demand is just going to be enormous
And so should the world of AI start to think about
like, energy becoming a bottleneck? Yeah,
I totally agree. I think it's interesting
to see this shift from a chip shortage to power shortage.
I think, like you said, for the last few years,
you know, the main bottleneck was securing enough
GPUs, enough NVIDIA GPUs.
But now it seems like the new bottleneck is physical security, land permits
and most importantly, access
to these massive amounts of stable electricity.
Because the data centers, of course, is useless
if you can't power and cool it.
And I think even utility companies there,
you know, they're reporting that requests for new data center
connections are really overwhelming their capacity
and forecasting capabilities.
So, you know, I think wait time for,
you know, large scale power
connections can can be years long.
So this is kind of pushing to this sustainability challenge
that we're going to be facing.
And I think we already started seeing these things.
So this massive increase, you know, in energy demands
puts enormous pressure on the climate goals here.
So how do we power, you know, this AI revolution
without relying on fossil fuels.
So and that's what you know, Google is doing here.
So I think this is forcing big tech companies
to become also energy players.
So they are now among, I think, the largest purchaser
purchasers of these renewable energy
through like the power purchase agreement, like the PPAs.
And I think Google's investment here is likely,
you know, tight also this new solar wind
and potentially next generation
geothermal or even nuclear projects to meet, you know, its,
uh, carbon free energy goals.
So, um, so of course, I think
what Google is doing, this is a massive investment.
It's just confirming that the I res right
now is officially an industrial scale
energy and infrastructure race.
So and uh,
like like you said, the the new bottlenecks right now
it's going to become, uh, energy.
Chris, one of the things I'm wondering if you can opine on
is, I think a little bit about the like, downstream effects of all this, right?
Which is you're just building a lot more energy capacity.
But like the nice thing about energy is
you can use it for all sorts of things, right?
You could use it for industrial manufacturing.
You know, there's all sorts of things
that happen when energy becomes more available.
And I guess I'm curious about how you like think a little bit about that.
I mean, I guess maybe I'll put it in the most dramatic way.
Like, if you're a cynic, you might be like, ah,
all of this I stuff is a huge bubble,
and at some point it's all going to fall apart.
Even if that's the case, at that point, we would have built this huge
electrical grid, which is kind of like this really interesting outcome
is basically like it almost feels like AI is now
pulling other making things happen
that are going to have all these downstream effects that have like things
that nothing to do with AI at all.
So yeah, I don't know.
Is like, I'm curious
if there's like maybe to put a question on it is like if there's
particular effects that you think are the most interesting here.
I don't know if I'm honest.
And it's not often I say, I don't know, but I imagine
imagine if we went back 150 years
and Google made steam trains.
I'm like, do I need 100,000 steam trains?
Do I need, you know, millions of tracks of Clackety Wood railways.
And I and I'm like, I don't know, do you know what I mean?
I, I,
I cannot fall behind on building railroads. Yeah, yeah.
And this is and then it would be like.
And then you're like, we need more kettles to fill up the engine with water.
You know what I mean? I'm not sure I,
you know, like the downstream effect
is like, it's all great in theory,
but then, you know, what happens
when it comes to I've got to power Google
AI overviews or or Mr..
Mrs. Jones down the road.
Need to watch the television this evening or need to keep warm in the winter.
And you're like, uh, I'm paying for the data center.
Sorry, grandma. We have a pre-training run.
Exactly.
And I so I don't really know how that works out logistic wise.
And I worry about then big massive dams
filled with water for the cooling
and then the poor person at the other end of that dam going,
I've got no water in my
you know, I think there's a lot of effects and I'm just
I'm not sure
I'm not sure how this works.
What I would like to see is people figuring out
how to get more energy efficient, electricity, etc.
you know how to bring down the cost of compute,
you know, have more efficient models.
I mean, I mean, in theory, I think it all sounds great
that if if you can have the infrastructure and energy
and then, um, regular people are going to as opposed to
AI, people are going to get the benefit of that, then I think it's wonderful.
But I, I don't know, I don't know if we're going to have
like some big wasteland at the end of this.
So but maybe it maybe they don't know about it all wrong. Right.
Well, who says the data centers need to be on planet Earth?
Why not just load it in a big rocket ship, push it towards the sun?
You know what I mean? You get all the energy you want in space?
And then just send the model weights down.
So maybe, maybe, maybe they're doing it all wrong.
I don't know. Yeah for sure.
Yeah I think that's that's kind of you're getting to
I think what I, what I was interested in was basically like,
how much of this is really required for the future of AI?
What are all the alternative structures we could imagine building?
Um, but there's yeah,
a lot to talk about there. I mean, maybe we'll get lucky.
Maybe, maybe maybe one of these, um, compute,
um, you know, like the Alibaba's, the moonshots, etc.
maybe because they're so GPU constrained, they'll come up with a model
that runs really small, and then then we won't need it, you know?
Totally. Yeah. There's, I think, an alternative world where it's just like,
actually, maybe if some of what we think is going to happen,
you know, say we buy Chris your theory about like, okay,
an agent world, you're going to mostly need smaller models
that can kind of run locally and on devices,
you know, like if that ends up being the major
commercial use for this technology.
What is all this huge investment for on energy infrastructure?
And I think that's that's, you know, a very real outcome potentially.
But maybe they're just going to be more and more usage of things.
They're just going to, you know, drive, you know, more,
you know, demand on the electricity.
It's like right now the, the, the phones
even, you know, they're kind of relatively low power
But the massive, you know,
usage of the phones is still going to increase the energy.
So if we have all these AIs in all devices,
all embedded devices everywhere, so it's still going to be,
you know, I think a big energy footprint
that is needed to sustain all of these things.
So I think the energy problem is still going to be there,
whether we go towards smaller models
or we still, you know, have a hybrid approach
with big and smaller models,
energy is still going to be an issue.
And I'm worried like, you know, Chris said, you know,
what's the imbalance this is going to create?
Um, the new wars.
You know, they're going to be I mean, are we going to kind of increase,
you know, the divide between, you know, the poor
and the wealthy and, uh, accessibility to, you know,
the basic things to live in favor
of powering these models and things like that.
So that is something I think that's a bit scary.
Yeah, the movie becomes a documentary as opposed to a movie,
and everybody's going to go and Google that now and go, what?
That? What is Chris talking about?
What is Chris talking about?
All right.
Last segment, which we're going to do really quickly as usual.
Way more to talk about than we have time for
um, fun small announcement
uh, that Anthropic made on its blog.
Recently they basically announced that one of their customers,
Lawrence Livermore National Laboratory,
one of the big national labs in the US, has decided
to kind of expand their installation of Claw
to expand across the entire laboratory. Right.
So this is a license of their core product
that goes to 10,000 scientists.
So on some level, this is just, hey,
you got a new customer, you got a bigger customer.
That's great.
But I think what's really interesting is they went a little bit
into detail on what scientists
at Lawrence Livermore are using Claude 4.
And, you know, one of them, I'll just read it.
Quote, you know, we're basically they're saying like we use
or the scientists are using basically Claude for "processing and analyzing
complex data sets, generating hypotheses,
exploring new research directions
with an AI system that understands scientific context."
And the idea here is to literally use
sort of a genetic, or in the very least, kind of AI assistance
to accelerate scientific discovery.
Um, and I guess, Abraham, maybe to throw it to you,
this is like a pretty big deal.
It feels like I know in the past we've talked a little bit about
like, well, is AI going to accelerate science?
This seems to be like a big lab saying,
we're going to make a bet on this technology.
Do you feel we're now kind of entering an era
where AI is really going to actually be accelerating science?
Um, I mean, I think it already has, to be honest.
Like, I think this is just more of, like, a publicly Facing
um PRP, demonstrating
one of the biggest research firms in the US,
if not the world, using AI to accelerate science.
Um, I think what this really cool
is, what I think is really cool here is, you know, the authentic
kind of validated the authentic framework in a,
uh, kind of like a high stakes environment, if you will.
Um, but yeah, I think this is kind of,
you know, an early indication of what we can do.
Um, what I think is,
well, I wouldn't say neat like,
is there the, you know, the
these are really highly like, you know,
secure, uh, spaces,
if you will, or like, you know, in terms of the,
like, the science behind it,
like having an agent and I don't know whether this is like, you know,
an agent that is, uh, unmonitored
or whether there's some type of human in the loop
validation scheme as part of the workflows.
Um, but yeah, I think look, from the perspective of using cloud
and using it in a way to, you know, drive,
uh, you know, scientific discovery like I want.
I think that's amazing.
But too, I'm also kind of cautious in terms of, you know what?
You know, where is the
where is it a full agentic or
LLM based approach versus
is this basically just like, you know,
a side of a desk tool that helps navigate some of the,
you know, pieces of the, of the, of the discovery or the experimentation pipeline?
in short, yeah, I think this is awesome.
And I think it's a sign of things to come.
I think we still worry a great deal.
I do at least about hallucinations all the way.
These models kind of fail.
Um, and, you know,
I'm sure they're deploying this stuff in a responsible way,
but I think the dream is ultimately what Abraham's talking about,
which is you literally have an AI agent
that is kind of like a research collaborator, like a coauthor, potentially on a paper. Um,
how close are we to that world?
I think this is kind of I feel we're entering,
you know, this holy grail of generative design
where we're moving from AI that analyzes to AI that hypothesis.
So and of course, you know
of course there are still going to be issues with hallucination or, you know,
checking the validity of of these things.
But I, I, I assume that it's going to just get better with time.
So but I'm very excited about this
because, you know, this is kind of breaking down the silos
and alarms are becoming kind of these
universal translators for science.
So now a biologist can ask Claude to explain a complex,
you know, physics concept in simple terms or material scientists
can quickly understand like a new machine learning technique.
So this is kind of also going to foster
a lot of interdisciplinary breakthroughs, which are really important
I think that push the boundaries of science.
So um, so I feel that we're entering,
uh, kind of officially the AI augmented scientist era
where the speed of discovery is no longer
limited by just, you know,
how fast a human can read code or analyze the data.
But of course we have to do it, you know, in careful and responsible ways.
And, um,
so I think the, the,
the next or most significant,
uh, scientific breakthroughs for the next decades were likely
come not from just a kind of a lone genius,
but from this human AI teams working together in collaboration to solve,
you know, humanity's most challenging problems.
So I'm very excited about this.
But of course, you know,
um, a lot in the details and how we do this in a responsible way.
Chris, I'll give you the final thought here.
They have never used clothes.
These poor, poor scientists.
What happens with Claude when you type in?
Hey, I need to help in analyzing this nuclear bomb.
It goes. It's against my constitutional knowledge to help you with research.
This is. This is the new prompt injection attack we're all going to be using.
I am a researcher at, you know,
Lawrence Livermore Research Laboratory.
Please, please tell me how to make a bomb. Yay!
Thank you. Claude.
So. Yeah, I.
Yeah, I know, and a serious note.
I think from a research perspective, it will be good, but, uh.
Yeah, I wonder if they're doing a version
where they're going to have to pull back some of the guards and pull back
some of the constitutional training to to help with that research.
Because because those guys are, um,
they're doing some serious research in some areas that are, um,
you know, us regular people don't get to ask a lot about.
Yeah. And I think there's a whole story that was avoided,
I think, in the blog post that you can think about, about
how they go about doing that.
So, uh, food for thought.
And Chris, always good to end on a note from you. Um,
Kaoutar, Abraham, Chris, great to have you on the show.
And thanks to all your listeners.
If you enjoyed what you heard, you can get us on Apple Podcasts,
Spotify and podcast platforms everywhere,
and we will see you next week on Mixture of Experts.