Vaibhav

[Vaibhav @vaibhavdaha]

Imagine companies just stop writing most of their backend code. instead of grinding on functions and algorithms forever, they fine-tune small llms for the actual jobs.

When a user does something, the model looks at it and decides what happens next no hardcoded logic telling it what to do. all that’s left is tiny glue so the models can talk to each other and get the work done. no real codebase, just the fine-tuned llms running the company.

Right now llms are writing the code humans used to write. In the future they mostly just spend time fine-tuning llms for whatever the company needs.

[Brian Armstrong @brian_armstrong]

We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case).

I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves.

We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them:

"/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started"

I imagine we'll have some backend services that could be replaced entirely by this new architecture.

[Vaibhav @vaibhavdaha]

Correct, and it’s not just about one model. Sometimes, you have to test across two models. It’s an iterative process where the same task goes through multiple rounds of user testing, which is challenging, especially as beating the current model is getting harder.

[john rho @john7rho]

Succeeding in selling data/envs is hard without VC because the lab promises you $ Xm in a few months to give data for them right now. But you have to evaluate the data you make by testing on their frontier models (Fable, Astra are super pricey). And you have to pay the market rate API cost while they make 80% margin off the rollouts you’re running for them (isn’t this kind of funny?). And if the data is bad they don’t have to pay you for the data, so you’re fronting the costs while also putting in a ton of manpower.

That three-month period when you’re broke and at the whim of one of a few AI labs is not easy, but exchanging some equity for upfront capital for tokens and compute makes it slightly more doable. Respect to the founders who make it happen.

[Vaibhav @vaibhavdaha]

@gokulr few are doing the same thing you just said, and they’re moving so fast.

[Vaibhav @vaibhavdaha]

This is true. Closed frontier labs have huge margins per token, so I don’t think they’re too worried about open source right now. They have enough room to keep investing and can continue to maintain healthy margins, while smartly giving things away through different incentives.

But things are moving quickly. Enterprises are building their own harnesses and evals around open-source models because of the lower cost, and they’re getting good enough results for their needs. These models are improving rapidly, while closed labs remain focused on competing with one another.

The real question is what happens if closed frontier labs decide to lower their prices to nearly match the cost of open-source models while still maintaining healthy margins?

[Vaibhav @vaibhavdaha]

Also because of long-term security concerns? Maybe possible if there’s reliable way to ensure the service can’t be used against India or misused by any government against India, while still meeting the other requirements and national policy.

[Elon Musk @elonmusk]

Starlink will be of great help to areas of India that have bad or no Internet and even no mobile coverage.

No country has perfect coverage, including America, but Starlink can fill in the coverage gaps for those who most need it.

[Vaibhav @vaibhavdaha]

I wonder what they’re going through right now.

[Gokul Rajaram @gokulr]

Fastest company ever to $1b annualized revenue

Nope it’s not Higgsfield or Cursor or Cognition.

Met a data company yesterday that did 0 revenue in November 2025 and 80m “in month revenue” in September 2026. They will hit $1b annualized from 11 months of being started. They are still under the radar so I can’t share who they are. Hopefully soon, with their permission.

I don’t make the rules.

[Vaibhav @vaibhavdaha]

@1kunalbahl Will be live in a few hours. Will DM you.

[Vaibhav @vaibhavdaha]

It is really hard to compete with frontier labs.
Building the best industry-specific harness will win.

[Elon Musk @elonmusk]

Important note regarding Grok @Bot:

Going forward, @SpaceX will use the best back end model for any given task, including Claude Opus 5.5, MidJourney, Suno and other leading APIs.

Whatever is most likely to give you the best outcome.

[Vaibhav @vaibhavdaha]

Codex really good at computer use.

[Yuchen Jin @Yuchenj_UW]

@bcherny I haven’t tried Claude Code Desktop in a few days.

2 things Codex does better based on my experience:

> better computer use. I use it to directly operate my computer for things like reimbursements.
> easier to work with oss models, and generally less verbose.

[Ilya Sutskever @ilyasut]

Neoclouds have limited cybersecurity. Next time agents successfully go rouge, they'll try taking over a neocloud to run more copies. This is bad.

Thus: neoclouds should greatly strengthen their cybersecurity and every company with strong cyber models should help with that.

[Nikesh Arora @nikesharora]

In two years from now you will be able to buy a neocloud for less than they raise at today. Just saying.

[Simon Willison @simonw]

Here’s my review of Qwen 3.8 27B - I can't remember the last time I've had this much fun playing with a local model that runs on my own computers https://t.co/iFMWjc8qle

[Vaibhav @vaibhavdaha]

Happy 80th Independence Day! ����🇳

[Vaibhav @vaibhavdaha]

@protosphinx Burn the boats.

[DAN KOE @thedankoe]

https://t.co/0XEnmV18FZ

[Tejeshwi Sharma 🇮🇳 @tejeshwi_sharma]

India’s electorate in 2029 by generation:

Gen Z: 350M (34%)
Millennials: 335M (32%)
Gen X: 225M (22%)
Boomers: 120M (12%)

Gen Z will become the largest electoral force for the first time.

The striking part isn’t the ranking. It’s that Gen Z is the only generation that will grow - electorate will add 60M between 2024 and 2029. Gen Z will add 120M. Every other generation will decline by 60M.

Moment of generation change is arriving !!

[Vaibhav @vaibhavdaha]

@nikesharora I can assure you that this is real because I saw you on TV twice.

[Vaibhav @vaibhavdaha]

People underestimate the data market's progress.

As models improve, they require different data, not just more of it. They need different data. Each lab needs something unique, tailored to its specific frontier model gaps.

Even within the same domain, one model demands basic coverage, another rare edge cases, and a third RL tasks that push it to its limits. So the same dataset won’t be equally useful to everyone.

Models create more demand for high-quality data as they improve.

Labs compete for providers who can deliver the right expert workflows, edge cases, evaluations, and RL tasks for their current frontier model.

Data is a moving target, not a commodity.

[will depue @willdepue]

A Stargate for Data

Labs are on a trajectory towards >$100B/year of data spend by 2030. As we begin the trillion-dollar compute project, we need to think about the equivalent civilizational-scale effort for the other core ingredient: data.

At the foundation of the scaling revolution is a simple empirical law: deep neural networks improve smoothly, near magically, as you scale two things in proportion — (1) the size of the model and (2) the amount of data you train on. And despite the scaling laws being brutally diminishing, we’ve successfully bitten the bullet of logarithmic scaling with exponentially larger clusters and datasets, and received incredible new capabilities in return.

But this exponential scaling is bound to hit some limits. Oddly enough, compute has compounded fairly smoothly without limit, with trillions flowing into hypercluster buildout. Instead, we’re starting to hit the limits of an exponential demand for data. Gone are the days of being purely in the compute-limited regime, where we had effectively infinite internet data but never enough GPUs, we’re now entering a data-limited regime.

Luckily, this limitation is coinciding with staggering improvements in AI capabilities. Incredibly, we seem to have a real line of sight towards automating a majority of knowledge work with the methods we have today. RL + pretraining, and the data for each, will be generally sufficient to achieve most economically valuable tasks, given some minimal algorithmic progress and continued compute scaling.

In a data-limited world, economic progress & scientific acceleration will be directly bottlenecked by our coverage in each domain. We need to see data collection as imperative, deserving the same civilizational ambition we’ve given compute.

The internet as a one-time subsidy

It’s underrated how much all progress in AI owes everything to the blessing of the internet, this one-time civilizational subsidy to deep learning, decades of unintentional accumulation of a perfect dataset: every book, blog post, image, video, paper, discussion, etc. all digitized and freely available. Without the internet, we’d likely see comparably minimal progress in AI today, and in fact, if you notice where systems currently underperform, it’s almost always a domain where web coverage is limited and data is private, expensive, non-digitized, or non-existent.

But we’re running out of it. There are only about 300 trillion tokens of useful public human text, and the internet doesn’t produce nearly enough new high-quality data to match what scaling demands — we’re soon to hit the limits of public data for pretraining. And though the advent of RL bought us reprieve — chain-of-thought RL needed a new form of untapped data, gradable math & coding tasks, also available online — we’re quickly running dry of hard tasks for RL as well.

Why do we need so much data anyways? Humans learn comparably in far less time, needing just one textbook where language models might need the equivalent of hundreds to learn a new topic. It’s possible we discover methods that are massively more data efficient — synthetic data, data efficient architectures, other exotic algorithms — but fundamental progress is slow and highly unpredictable, and the recipe we have just works today.

And, while I’m wary of getting too deep here, even arbitrary data efficiency can’t replace data that just doesn’t exist in the first place. There’s a massive amount of missing information on the web: the dark matter of the internet — tacit knowledge, undocumented processes, etc. — most of which was never published and lives only inside organizations, the physical world, or just in people’s heads. I’ll leave it here and say, for reasons far longer than I can fit in this post [1], it’s best to operate on the assumption that our insatiable desire for data will continue as it has for the last decade.

There will be >$100B/year in data spend by 2030

We’re not screwed yet, of course. Only a fraction of useful data in the world is on the public internet, the rest is stored inside private datasets, corporations, personal archives, universities, governments, and otherwise. Labs can and will continue to license these private datasets, or create them from scratch, like Anthropic’s book scanning project. And we’ll increasingly task human experts to manufacture new high-quality data, with a large fraction of hard RL training tasks already being sourced this way.

But collecting this data, unlike before, will be expensive. As the free internet dries up and demand for data rises, we should see labs investing equally in data as compute, likely spending a significant fraction of their compute budgets on data. As we see trillions spent on compute, we should also expect hundreds of billions spent on data (human data & collection budgets), given their equivalent importance. And, notably, data spend is already tracking this way: total data spend across vendors, not counting internal lab efforts, is already roughly $7 billion per year. It’s quite reasonable we’ll see >10x by 2030.

Data is the moat

Data becoming increasingly private will also majorly shift the competitive landscape. While compute is a commodity — everyone buys the same chips and builds the same clusters �� data really isn’t. The big reason why frontier models have felt eerily similar to one another, until now, is they were trained on substantially the same internet (pretraining data variability across labs seems pretty low). As labs diverge onto more exclusive, manually collected corpora, I think models will begin to increasingly diverge.

OpenAI pulling ahead in mathematics and Anthropic in cybersecurity isn’t an accident. I really think laser-focused collection of high-quality midtraining tokens, custom RL tasks, environments, with dedicated research effort, has driven much of the visible progress in the last year. James Betker has an excellent blog about “the ‘it’ in a model is the dataset”: model architecture and compute buy you efficiency and order-of-magnitude performance, but ultimately, models, of any architecture, are such incredible approximators of their dataset that the core meat of a model boils down to just that, nothing else. Data is a major moat.

AGI long, ASI short

As I’ve tweeted before, I’m confident that, despite the narrative, the data labeling industry will continue to fuel great businesses and be an excellent AGI long, ASI short. The argument is just: By the time the AGI labs no longer need data, it’s probably over for everything else too [2]. In this frame, the last companies left should be the data companies, as the last speck of economically relevant data is sucked in. And these companies are already among some of the fastest-growing companies in history: Mercor, founded three years ago, is rumored to be doing $2 billion in revenue with something like a few million expert labelers under contract.

While these businesses are very non-stationary, what type of data is needed shifts constantly, I don’t think that diminishes their value. The long-tail of the economy is long, and the value isn’t diminishing as you extend farther into more obscure information: as models get more capable, the value of the marginal dataset goes up, not down. Automating a full job means covering its full distribution of tasks, tools, edge-cases, and long-horizon loops. There’s some O-ring logic to it: a dataset that buys a 1% bump can justify a previously unjustifiable collection cost when it’s the difference between a system that does 99% of a job and one that does all of it [3].

The competitive dynamics of the data industry are still evolving but as demand for data is increasingly niche, ultra high-quality, expert-generated, I think we’ll see real consolidation. Again, contra-narrative, we’ll probably see true competitive differentiation built on brand, quality control of data (which, from personal experience, can vary massively), as well as in network effects from the talent networks themselves over time. We’ve already seen rapidly shifting data type demand work in favor of incumbents, benefiting those with early knowledge of where the market is headed.

The binding constraint

It’s truly remarkable that we seem to have the recipe — pretraining + RL — to absorb most economically valuable work, despite being far from a lot of what we expected from “AGI”. The same way chess engines revealed we never needed general intelligence to solve chess, as we originally thought, we’ll soon realize that software, mathematics, and the vast majority of the economy (including physical, just running ~3 years behind!) are the same. If recursive self-improvement or some other algorithmic breakthrough arrives, that’s wonderful, but we really don’t have to wait for it. The binding constraint between here and an automated economy isn’t that, it’s data coverage: every app, workflow, edge case, process, etc. sitting in private stores or someone’s head.

Ultimately, while we make tremendous strides in more efficient model architectures, and clusters like Stargate equip us with zettaflop-scale compute, we really aren’t making rapid progress collecting the data we lack.

We’ll soon live in a world where we have the methods & compute to accelerate scientific progress or economic growth, but not the data. And we’re already there today: frontier models would surely be as good at accounting/many medical tasks/legal advice as they are at software engineering if we only had the same pretraining & RL coverage as we did for code.

I really want to drill this in: The speed at which we automate the economy is going to be directly rate-limited by our ability to collect data about it.

Worth noting that under this assumption, with data as defensible and directly proportional to economic & scientific progress, data should also be considered a national strategic asset like compute. Imagine what we’d do in a world where we had a Manhattan Project-effort for AI and needed to mobilize data collection as a limiting factor. We should be concerned about China, with greater state capacity and authoritarian economic control, being capable of mobilizing data collection at national scale, potentially compounding their economy and scientific output faster than us down the line.

A Stargate for data

I’m leaving my complete ideas for a future post, as this one is already far too long, so I’d really like to pose the question here. Stargate exists because we organized trillions of dollars, international strategy, gigawatts around compute as a fundamental ingredient. What would equivalent ambition look like for data?

Obviously, scaling data collection, a heterogeneous mass of information across the economy, isn’t going to be as clear as scaling compute, as a homogenous infrastructural effort. A core division will be first, coverage — all uncaptured knowledge sitting across the economy/science/physical world and all that simply isn’t recorded — and, secondly, sheer volume in the domains we already train on: more hard math tasks, more high-quality web text, way more coding data, more legal drafts, etc.

I have a post coming soon which breaks down my proposals. There’s a lot of room for creativity. Quickly, we’ll probably want to start with a deep census of what we have and what we’re missing, predict what the 2030 model will still be bad at and work backward to what we should be collecting today. You can probably license a large amount, leveraging high lab valuations to buy datasets or companies altogether. There’s an adversarial nature to a lot of this collection with firms, so there’s lots of engineering to do this correctly. We should go convince important companies to turn off deletion policies, even if we’re not buying from them yet. Data flywheels in consumer products will be massive. Confidential training, government legislation for grant-funded research, running companies at a loss for their data, etc.

We’re headed towards hundreds of billions in expenditure, national prioritization, and major data limitation on the horizon. We have a great opportunity to think creatively about what a megaproject for data would look like: How do we, deliberately this time, construct the next internet’s worth of data?

Footnotes:

[1]: I’ll probably soon publish my much longer post explaining my position on data efficiency and why the value of this data is still pretty high in most worlds regardless of new algorithms.

[2]: The “AGI freeroll” bet: heads you win, tails ASI flips the world upside down anyways.

[3]: We already see a glint of validation of this point, given the data market is strongly tilting towards ultra-high-quality agentic data, rather than unskilled labeling — niche expert workflows, live environments, and evaluations requiring increasingly obscure talent & knowledge — yet shows increasing, not decreasing, revenues.

[Vaibhav @vaibhavdaha]

High-quality data and high-quality internal benchmarks are interdependent. You can’t have strong progress with only one of them being good.

[Deedy @deedydas]

“The dirty secret in AI is that everything is a data and an eval problem.

The best models have the best data and best internal benchmarks. The mid ones buy a lot of data, not the best, and hillclimb public benchmarks.

(you need a lot of compute too)”

– Stanford CS Professor

[Vaibhav @vaibhavdaha]

The biggest misconception in AI is that better models will eventually need less high-quality data and labeling.

In reality, the opposite is true.
Data and label quality set a hard ceiling on what any model can achieve. no amount of model improvement can overcome noisy, inconsistent, or poorly structured data. As models get more powerful, they become even more sensitive to the quality of what they’re trained on.

They don’t just need more data; they need much better data with clear meaning and relationships. Getting that level of quality is extremely difficult and expensive.This scarcity is already forcing companies to treat data operations as core work.

We’re seeing large teams of engineers being reassigned to full-time data labeling and human feedback roles internally.

Expect even more organizations to shift focus inward and build stronger in-house capabilities rather than relying solely on external providers.

[Yun-Ta Tsai @yunta_tsai]

Many people think any given ML project is 99% training.

In reality, it’s 50% evaluation, 40% data cleaning, 8% integration, and 2% training.

The first two set the noise floor for learning. No ML magic matters; the model cannot lower the noise floor, as that’s the optimal bound of Shannon encoding of your data.

Thus, not a single day goes by without me thinking about ontology. Even the old labels have to be constantly reviewed.