Anthropic-OpenAI-Led ITpocalypse: Inside-out & Opportunities

Turning Market Panic into Enterprise AI Opportunity

Anthropic-OpenAI-Led ITpocalypse,” wiped out $285–300 billion in software stock value. Vinod Khosla mentioned IT services will disappear by 2030. Products like Salesforce, SAP also have been bleeding.

But it opens opportunites beyond obvious. Understanding deeper nuances helps us to catch them & inflexion points beyond obvious. A stack similar to OpenAI/Anthropic for mid-market & vertical focus will be so much invaluable.

What’s underneath the stack, what’s the moat; what changed drastically & what are downstream opportunities. We cover these & similarties with our own stack at Vaayu & technical papers

Anhtropic spent 58M per 1M earned in year 1

But now there’s a shift towards new regime. Dust stelling where smaller players can compete with much less capital We cover inside OpenAI/Tech stack, how to build it for smaller vertical players like us & what are opportunities.

None of the latest releases from Anthropic or OpenAI are “frontier AI” in the pure research sense. These are more of Engineering led releases. Below table fo Anthropic’s revenue vs funding shows the strategic shift.

Year	Annual Revenue	Cumulative Revenue	Funding Raised in Year	Cumulative Funding	Cumulative Funding ÷ Cumulative Revenue
2022	~$10M	~$10M	~$580M	~$580M	~58×
2023	~$200M	~$210M	~$5B	~$5.6B	~27×
2024	~$1B	~$1.21B	~$3.5B	~$9.1B	~7.5×
2025	~$13B	~$14.2B	~$13B	~$22.1B	~1.6×
2026*	~$16B (mid-estimate)	~$30.2B	~$30B	~$52.1B	~1.7×

Recent releases by Anhtropic are more of engineering work than paradigm shifts of tech

More recently they are focusing on revenue. So more engineering work than paradigm research. Like Opus increased the context window enabling better Multi step reasoning. OpenAI releases to are enginerring focused than erswhile pardigm shifts improving benchmarks. While the mainstream media covered only surface level stuff – we cover deeper look.

Inside Anthropic & OpenAI stack & opportunities for smaller players

Closest explaination of changing stack comes from Gokul Rajaram, our advisor.

Our chat triggerred from his X post where he argues “It’s one of the clearest representations of how AI companies plan to build next-gen systems of action on top of existing SoR, and why the markets are so worried about the future of software companies”

I had a brief whatsapp exchange with him about Vaayu’s market positioning as Digital workers with Decision Intelligence layer for Cleantech & Cooperative banks.

Also Anthropic’s release of Opus offers higher context window, which allows better multi step reasoning.

Enterprise AI Stack & Stakes will change drastically

So it’s important to understand the mechanics in detail, what’s happening & why.

It provides us to preapre to hit or at least aim one the inflexion points.

At Vaayu, we always envisioned enterprises run by AI, Supervised by humans.

Why it matters – ITPocalypse & New Wealth concentration

I have long maintained that AI will cause a lots of new wealth generation & concentration of wealth & impact at every level.

Eg. Anthropic valuation has reached $380B in 5 years while Infosys is at $60B in 45 years. But it’s also true at smaller level – startups, SMEs, Businesses, Individuals. Power law will become more dominant.

Here are some mechanics why there are 10s of paths & shots being buiilt on this path. As an AI Company we just have to hit 1.

Every sector will need a new Enterprise AI stack or will perish

Downstream opportunity for mid-market play

While fortune 500 will have access to this emerging stack, mid market is underserved. If you think just Anthropic claims or social media beliefs are true. Read below.

Anthropic’s Exec claim 100% of code is now written by AI, but are hiring engineers themselves.

It’s not easy to build deep Enterprsie AI stack in true sense. Most VCs or founders or even techies seem to care about workflows or use cases. But the real difference is underneath.

Achal – my co-founder & myself have done large scale continet migrations & we know how big chaos this is going to be at even Fortune 500.

The mid-market & vertical will remain underserved. Nor they have elaborate teams to build it.
There are 200-900 applications that large enterprises use. These though look similar at surface, there are at least 2000 ways to solve the same problem where lots of money is paid.
Merely Anthropic & OpenAI owning the stack & even giving APIs doesn’t mean it will be available to mid market
It needs niche engineers still to build such stack & complex execution

Enterprise AI Stacks & Stakes are going to change.

A stack similar to OpenAI for mid-market & vertical focus isn’t that easy to build

Vaayu’s focus has been always deep proprietary stack. We focus on rather some aspects which are based on research. Full stock approach.

Our theme has always been this stack where autonmous AI Agents run the enterprise supervised by humans. Our vision stemmed from our jobs in 2019-2022 in Autonomous driving. Achal & I then converged this to Enterprise AI which has been most of our career work.

Our Engineering-Led-Applied-Deeptech-AI approach

Anthropic & OpenAI shift

New age Enterprise AI stack has to emerge, & it opens new downstream opportunities

When it’s not poetic research in true sense & Engineering-Led-Applied research, new aveneue open for smaller players. You no longer need Stanford AI Labs grads, you don’t need 500 millions.

This is the approach – Vaayu has always endorsed. It’s less capital intensive, complex enough to be defensible in the niches, needs Research & deeptech – but not very very long period.

I have long been evangelizing, writing & following this approach at Vaayu. This is novel but not new. Many major Fortune 500 companies us this approach in niche pockets – JPMC, Citi Bank, Salesforce, Oracle, Credit Suisse, BMW, Mercedes. All have such pockets of deep & niche work

This Achal & I have oberved closely at our jobs at Fortune 500 working with deep & niche depatments where PhD researchers from Standford, MIT, Oxformd & Munich technical universities work at Fortune 500.

Multiple opportunities for such a stack – we just have to hit 1

Mind you OpenAI stack that I showed has only partial picture.

Each of 3 pink layers of OpenAI Frontier are budding ground of innovation. Eg Business Context. Each consists of multiple subcomponents which of course they won’t show. Each of them is budding ground for inflexion points. There are multiple ways to build this stack. Vaayu’s stack is diffrent but principally – there will be different stack for different sector conceptually looking similar to below

Eg, below is our version of stack for banking presented long ago at Citi Bank. Where the core represents above stack & can power entire banks. Of course we start with some specific niche to begine with.

Eg. Vaayu is building modern Enterprise AI on one of the paths. Add to it sectoral & customer segment complexities. Add to it – there are many substreams & ways to solve the problem. Just one of factors in stack can give unique niche edge.

Like we build Digital workers for Cleantech. With Decision Intelligence for GTM & Profitability

But within that we focus on Lead-to-revenu & profitablity via cashflow arbitrages. For specific sectors – EV, Solar, Fleet managers, sustainbaility ecommoerce, EPCs & more. Plus we work on latent states for decision intelligence. That’s broad enough & specific & closed enough as a cohesive focus for uniqueness.

Multiple inflexion points with Engineering-Led-Applied-Deeptech

Another point we argue is though VCs like it – putting all bets on single compoent will need huge capital like Anhtropics example. So it’s better to be full stack but focus on 1 key differentiation. Merely context graphs aren’t gonna work or are waay too risky.

Take eg this.

Last time when Context Graphs from Jaya & Ashu made headline – I wrote that Context graphs aren’t sufficient in business goals achievements, nor they are the most advanced tech to achieve same. Graph tech has existed for long & we have proof of that it’s not sufficient. For companies built on past decision may be suboptimal that what’s possible with new age.

So a full stack Enterprise AI stack will have multiple budding ground to hit a big dimension with slight improvment focused on particular sector. Relying everying on context graph is very risky since other compoents of new stacks may not emerge at same speed of investments done based on virality.

There are several components witin each layer. And there are 10s of variations needed for sectors. Which are supposed to be built by Infosys, TCS etc for large companies.

Vaayu’s own work is around AI Digital workers, Small Reasoning Models with Decision Intelligence for Cleantech & Small finance banks.

Every enterprise willl change – How to capitalize on ITpocalypse

So basically the stacks are going to change. someone still has to build it. Very few actually can in true sense.

Vaayu – Digital workers with Decision Intelligence layer for Cleantech & Cooperative banks.

Mid market focus for Cleantech & Cooperative Banks
Vertical stack
Backed by Research.

In next posts, We cover market shift, inside of the OpenAI & Anthropic stacks & then introduce our technical Research paper & path foward how a path to Enterprise AI may look like.

Mind you, AI, Software & Tech – There are 10s of paths to solve the same problem. Our tech paper introduces some novel paths.

Capital it takes

I have also been talking to more VCs, Product leaders, AI researchers & 10s of customers as well.

Anthropic spent 580M to earn 10M in year 2022. This was when frontier models were taking shape. Recent product releases are more engineering focused. This approach takes $1-$3M to earn $1-$2M. Based on how deep you want to buld.

Summary & next steps

Vaayu builds Digital Co-workers with Decision Intelligence for Cleantech & Banking. Full stack contains Small Resoning model, deep proprietary stack & data engine. You may consider it an advanced version of Salesforce agentforce with propritary stack, IP & unqiue sectoral focus.

If you are PhD, or US firm, Investor or Salesforce/Zoho/Microsoft partner or PwC/BCG/Deloitte partner – Do reach for collabs & to jam. There are unique opportunities for collab.

Read Vaayu’s overlapping work on Digital Co-workers with Decision intelligence – Technical papers

If you prefer more techncial style & depth – pls request here.