Through most of his professional life, first as an entrepreneur and then as a venture capitalist, Sateesh Andra has almost always been able to decode technology cycles. But AI, he says, is different. The rules that defined prior cycles no longer apply. And that makes it difficult to visualise which playbooks will define this era.
Andra co-founded Endiya Partners in 2015 out of Hyderabad, at a time when the city was not yet the startup hub it has since become, and when backing India-born deep tech and enterprise software startups was still considered unusual. Across three funds and a portfolio of 30-plus companies, including Darwinbox, SigTuple Technologies and Kissht, the strategy has earned the venture capital firm outcomes with a consistency that’s a bit rare for a specialist technology investor in India.
In his first guest article for The Runway, Andra revisits the framework he articulated about 15 months ago about where India should stake its claim in AI. The question he keeps coming back to is bigger than any individual opportunity: can India win this cycle as an ecosystem?
Dive in.
India’s Right to Win in AI. Revisited
I have watched technology cycles since the mid-nineties. Something is happening in the current cycle that has no historical precedent.
The repercussions, for every layer of the technology stack and for every Indian company operating within it, will play out over the next few years in ways we cannot yet fully visualise.
It also forces a harder question than the one I asked about a year ago in my article The AI Gold Rush: Where Should Indian Startups Stake Their Claim? The question today is not whether India can win in AI but whether we can win as an ecosystem.
To understand what makes this moment different, it's important to track back to how we got here. The technology industry has reinvented itself roughly every decade. From 2000 onwards, we've moved from the client-server world to the internet and then on to mobile, the cloud and now AI.
Each reinvention has been defined by a single dominant question: what is the computer?
The first cycle belonged to the workstation. Sun Microsystems running Solaris. Silicon Graphics running Unix. Microsoft running Windows. The computer was a physical machine. Compute and operating system bundled together. Then the network became the computer, omnipresent, a shift Sun Microsystems first called out in the eighties, and Cisco helped make real. Amazon with AWS, Microsoft with Azure and Google changed the rules again. This time, the cloud was the computer. Infinite, on-demand and accessed over a network.
In every prior computing shift, the companies that built the foundational technology mostly stayed within their layer, or expanded into adjacent ones over years, not months. They built the base and let software vendors, implementation partners and application companies build businesses on top of it.
That's what created room for ecosystems to grow. A thousand companies could build viable businesses on top of a stable foundation.
Today, NVIDIA with Anthropic or OpenAI or Google Gemini are the computer and operating system. But, here's where it diverges from prior cycles.
Anthropic and OpenAI are going to Salesforce and saying: adopt me, use me as your AI layer. They are then going to TCS and saying: when you implement Salesforce for your clients, use me. But it doesn't stop there. Anthropic and OpenAI are also going direct to the end-customer, the enterprise, with their own services capability. At the same time, they are deepening partnerships with TCS, Accenture, Deloitte and others.
In a single motion, one technology company is capturing value as the software vendor, the implementation partner, and a direct competitor to the client's own internal capabilities, simultaneously. That, for me, is unfathomable. This time something profound is happening. And, it questions India's right to win as an ecosystem.
When I wrote that first article in April last year, we mapped where value was being created in AI into four buckets – trailblazers, innovators, amplifiers and users.
The Trailblazers are companies building the fundamental infrastructure – large language models, compute chips, inference chips, vector databases, and so on. These companies are concentrated within a 10-mile radius of San Francisco and require extraordinary capital. So, for example, Anthropic crossed a $50 billion run rate in mid-2026, even as it locked in $200 billion in compute spending with Google alone over five years.
That’s not India’s game, for the most part.
Then we have Innovators, which includes vertical AI application companies. These are AI-native from the ground up. This is the layer, where technology meets human beings directly, at which you typically derive the most value.
These are the companies that Endiya Partners is backing most actively. We have Maieutic, a kind of Cursor/Claude for chip design. Then there’s Perceptyne in physical intelligence for automotive and electronics manufacturing assembly lines, Matters in security, and ThirdAI in industrial predictive maintenance.
Under Amplifiers, we have two kinds. First, established software companies such as Salesforce or Darwinbox (from the Endiya portfolio) which are embedding AI into products that already sell to existing customers. Second is what I call services software. A new category that is structurally inverting the old BPO (business process outsourcing) model. Previously, services implied people implementing a process leveraging technology. Now it’s flipped. AI drives the process and workflow, with humans in the loop when needed. An inverted pyramid.
The last bucket is Users. That's everyone. Just as we all became internet users, we are all AI users today, whether it's for coding, customer support, marketing, sales, back-office, or research, in the enterprise scenario. It's faceless for the consumer but AI is at work behind the scenes.
The framework still applies. But the stakes have changed.
My belief is that good Indian companies, the truly AI-native ones, will go to half a million revenue per FTE (full-time-employee). In our own portfolio, I wouldn’t be surprised if teams of 20-25 people hit $10 million in revenue. For these companies, AI is not a product feature. It is the operating model.
Microsoft and Google used to benchmark at roughly a million dollars of revenue per FTE (full-time-employee). That number has moved up nearly 2x. VMware under Broadcom cut its workforce by more than half while pushing margins up sharply and targeting a doubling of EBITDA. AI-native companies such as Anthropic are reportedly already operating at close to $8-$9 million per FTE.
The big question for India is whether we can manifest success in AI as an ecosystem. Take SaaS as a reference point. How many real winners do we have even today? Maybe 15-20 over a decade. In AI, it isn’t clear yet what form the disruption will take. Will there be a long tail of small companies, each doing $10 million revenues or would you see consolidated players?
My honest answer – I don’t have clarity right now.
The views expressed in this article are solely those of the author and do not represent the views of The Runway, its editors, or any organisation the author is affiliated with. This article is intended for informational purposes only and does not constitute investment advice. The author may have direct or indirect interests in companies or sectors discussed.
Thank you for reading The Runway. We’d love to get your feedback. Drop us a line at hello@therunwaynews.com or just reply to this email.
The ecosystem question has a workforce answer, and right now the numbers say not yet.
Nomura's August 7 report counted 83,100 AI-related hires in India against 31,921 losses since 2022. Close to three hires per exit. But Nomura's own caveat is the real finding: displaced workers rarely move into AI engineering roles.
Naukri's July index shows the same split. White collar hiring up 5%, AI and ML roles up 33%.
So the AI layer is being built on top of the IT services generation. The people underneath it are not moving up into it.
An ecosystem only wins if that transfer happens.
Which Indian company has actually moved a support or testing team into the AI stack at real scale?
Zia. AI career strategist for Indian professionals. itszia.ai