Posted by 
DataBricks raised $5B at whopping $190B valuation. But that’s news. This blog is about what is in it for our readers- you, us, investors. How Vaayu World Models fit into the picture.
Pragmatic & low Capex Infra to build Vaayu Enterprise World Models
Though the number may look intimidating. But neither Databricks started by saying – “Hey! Let’s build a $200B company”. Nor it has to be the same journey always.
So, in this blog we cover prgamatic way of category leadership, when Enterprise AI infra is getting built.
We slice & dice in a way that a seed of $2m may be good to start in a niche, even to consolidate category leadership. We make a case that Atlan has reached $750-800m valuation, but didn’t raise Billions.
Similarly, Deepmind started with meager $2m cheque. Important point here is, identifying the emerging – that is, Decision Intelligence, World Models & understand nuances around it. To build a niche category in much less capex, possible from India (& later from San Jose) with World Models.
AI Decision Intelligence & infra on roll
From Databricks to World Models: Why Vaayu Is Building the Next Layer of Enterprise AI
For the last decade, one of the biggest transformations in enterprise technology happened underneath the application layer.
It was data.
Then, Decision Support Systems, Digital Twins & Ontologies were placed on it.
ChatGPT era led to Agents & Agentic flows.
Reasoning & Decisioning.
Recent raise by Databricks shows momentum for Infra, Decision Intelligence
But now the next main frontier is World Models
~1300 companies worldwide in overarching Decision Support Space
Basically, Data, Decision Intelligence, Decision Support, Context/Intelligence layers (Ref Tracxn)
Databricks is ranked 2nd, Palantir is ranked 1 with $350-$400B valuation.
Vaayu is ranked 228 among these 1300 startups worldwide merely at seed stage.
If you take the niche of world models among them, then there are roughly 20-40 companies worldwide working seriously on it, while most focus on Robotics, Autonomous cars & Humanoids.
Vaayu works for Enterprise World Model. This means capex requirements and model size are reduced drastically.
| Rank | Company | Category | Funding / Capital | Valuation / Market Cap | What it does |
|---|---|---|---|---|---|
| #1 | Palantir | đźź© Operational Intelligence / Decision System | ~$2.5B tracked funding | ~$383B market cap | Enterprise operational intelligence: connects data, ontology, business logic and operations to support high-value decisions. |
| #2 | Databricks | 🟦 Data Infrastructure + AI | $5B latest raise | $190B valuation | Data lakehouse, analytics and AI infrastructure; increasingly moving toward AI agents and applications. |
| #5–15 | Atlan | 🟦 Data / Context Infrastructure | $206M+ | $750M valuation | Data control plane / context layer connecting enterprise data, metadata, governance and AI systems. |
| #21–35 | Aily Labs | 🟩 Decision Intelligence | $100.6M | Not publicly disclosed | AI-driven decision intelligence across finance, operations, R&D and commercial decisions. |
| — | Altair | 🟪 Simulation / Computational Intelligence | Acquired by Siemens | ~$10.6B market cap / acquisition value | Simulation, digital twins, HPC, data science and AI for industrial and engineering decisions. |
| #144–228 | Vaayu | 🟩 Enterprise Decision Infrastructure / World Models | Private | Private | Enterprise World Models and Decision Intelligence: modeling enterprise state and dynamics to predict, simulate and optimize high-value decisions. |
Vaayu ranks 144-228 among ~1300 worldwide startups at the seed stage. (Ref Tracxndata & wehter you count direct & indirect competitors)
Databricks raise & World Models
The company has raised $5 billion at a $190 billion valuation, with annualized revenue above $7 billion, while expanding from its Lakehouse foundation into products such as Lakebase, Genie and Unity AI Gateway.
There is a bigger lesson here:
The largest enterprise technology opportunities often begin as deep technical infrastructure—and then move upward toward intelligence.
Vaayu World model fits well on top of Data Lakes such as by Databricks
Read more about use in Banks of Vaayu Enterprise World Models
Next Databricks won’t be the same
There is a bigger lesson here:
The largest enterprise technology opportunities often begin as deep technical infrastructure—and then move upward toward intelligence.
Vaayu believes the next layer could be World Models and Decision Intelligence for the enterprise.
Read more – Can this be an opportunity to build a mini Deepseek moment?
Positioning for Category leadership with Enterprise World Models
We have repeatedly maintained that AI provides unique slicing & dicing. Such that a single right hit can get strart ups in World charts. We call it Deepseek moment.
Following section explains how we aim for category leadership
Enterprise world Models rather than for Humanoids
World Models are one such aspect. While most of the world believes that they only have to use, tweak or fine-tune existing models released by Bigtech. Vaayu thinks differently.
We adapt technologies used by Tesla, BMW, Deepmind, and Meta for latent space representation learning & JEPA, Data fusion, Decisoin making, Dynamic system under chaos.decisiondynamic systems
Focus on Small Reasoning & Decision & world models
Small models specific to banks reduce cost, create proprietary infrastructure. Consider like BloombergGPT.
Read details about Vaayu JEPA & Small Reasoning Model
JEPA to reduce cost & sample efficient learnings
JEPA can improve sample efficiency by learning compact latent representations and predicting what matters in the future, rather than modelling every raw input detail.
This can reduce the compute, data and training cost of building World Models—especially when enterprise have a data lakes
Such World Model infra is on top of Databricks datalakes, Context & knowlege graph
JEPA & World Model for Banks in layman’s terms
Have you heard someone say – “This is how the world works nowadays”?
So, they are pointing to “Some hidden relationships, including 2nd-order & 3rd-order impact of data & systems, state of the world & how the world would react to your action (dynamics). Further Decisions & policy-making based on the above have to be tailored.
Imagine an AI model that can tell 2 months in advance whether a lending portfolio will default or a fraud will happen, or if payments will be delayed based on multiple parameters.
World Model Example for banks in layman’s terms:
Eg. Model can discover that in Vidarbha, a combination of extreme heat and rainfall variation doesn’t immediately affect loan repayments—but say UPI transactions start decreasing, which leads to lower gold purchases and delayed payments by jewellers in specific PIN codes, which can become an early signal of stress in a particular borrower segment.
USP of JEPA & World Model for Banks over Bank’s existing models & LLMs/SLMs
Existing bank models are usually built to solve specific problems—such as credit risk, fraud or collections—while LLMs/SLMs are primarily designed to provide generic Q&A.
World model & advanced techniques used by Vaayu are designed to understand the bank as a dynamic system: Uncovering hidden relationships, including 2nd-order & 3rd-order impact of data & systems, dynamic state relevant to a particular decision & how a particular decision will impact the final outcome
This enables use cases such as early fraud signals, emerging portfolio stress and systemic-risk prediction, rather than making only individual predictions.
Why JEPA & World Models for banks is an advanced niche technology over Databricks like Data lake & infra?
World Models are emerging as one of the more advanced and specialized directions in AI research beyond LLMs & SLMs.
Major AI labs such as Google DeepMind, Meta and NVIDIA have released models. While The field now has active research but very concentrated advanced programs across leading institutions including Tsinghua University, Stanford, UC Berkeley, MIT and Carnegie Mellon, alongside
The technology has also attracted significant attention from frontier AI companies and investors, but the number of researchers and companies working deeply on World Models, JEPA, latent-state modelling and predictive dynamics remains small compared with the much larger LLM ecosystem.
Read more about Vaayu’s advanced Decision intelligence – here
Vaayu’s approach: Applied Deeptech AI
We are deliberately not positioning this as frontier-model research competing with the largest AI labs.
Our approach is different.
We focus on Applied Deeptech AI:
Take an emerging technical frontier → build a specialized architecture → solve a high-value enterprise problem → deploy it into a real decision loop.
The objective is practical:
Research → IP → Applied AI → Enterprise Decision → Economic Impact → Eg reduce NPA by 5-10 BPS
The next Databricks may not look like Databricks
Databricks demonstrated how research in distributed computing and data systems could become a massive enterprise infrastructure company.
The next generation may emerge from a different research frontier:
World Models.
Predictive representations.
Latent-state reasoning.
Decision Intelligence.
Enterprise simulation.
The companies that commercialize these ideas may not sell another chatbot.
They may sell something much more fundamental:
A machine that understands how your business works.
And eventually:
A machine that can explore what happens before you make the decision.
That is the opportunity Vaayu is pursuing with Enterprise World Models and Decision Intelligence.
The thesis
Databricks built infrastructure to make enterprise data computable.
Decision intelligence then makes to uncover hidden signals & make high value decisions better with Vaayu Enterprise World Model based infra.