We are living in one of the most exciting times. Our lives are changing as never before right in front of our eyes. And as we all know, AI is what is driving this remarkable disruption.
For Pentathlon, with our early stage B2B Tech thesis, this disruption is a huge opportunity. Identifying investment opportunities where Tech (in the current context, AI) is disrupting businesses across industry verticals, is a problem of riches! Whether it is manufacturing or healthcare or banking and insurance or education, it is guaranteed that AI will disrupt the way they do business.
The AI opportunity
Obviously, we are not the only ones eyeing the opportunity. Huge investments are being made in the infra required for AI, as AI requires humongous amounts of data, computing and energy. For the GPU’s and the data centers, there is literally a capital-intensive “arms race” to remain ahead of the competition. Nvidia is now the most valuable company in history.
Tech giants are building LLMs – ChatGPT, Gemini, Llama, Grok, Claude – and as they compete, we benefit with each stride they take. The speed at which the world has taken to AI, sometimes it’s difficult to believe that ChatGPT came about only 2 years ago!
In the layers of industry applications, large funding at high valuation is being offered to “AI-First” companies led by techies, expected to bring in “generational” transformation of conventional industries. Over 50% of the global VC funding this year went to AI companies.
Already we have seen the amazingly fast growth of some AI companies, such as Cursor and Lovable in the coding space, which seems to justify the hype around AI. That is leading to investors and tech companies pouring even more money into building AI and others into using AI. The narrative is only getting stronger due to FOMO (Fear of Missing Out) – nobody wants to be left behind.
“Bubble” or Hype Curve
For the public, this is eerily like the internet bubble quarter of a century back.
GPU’s will diminish fast in value, as they get replaced by more powerful ones. Algorithms will use chips more and more efficiently, as Deepseek showed. Data centers built in a hurry will be very costly. Investment commitment in power for them is USD 1 Trillion (although power is fungible)! Building an over-capacity of these assets seems exactly following the dotcom boom and bust. Only, the investments are much bigger (17x!) and much more difficult to return.
LLMs are the new battle-field for the Tech giants. The incumbents have their never-ending cash piles, and the newer players are building their own through investors. A very high stake battle indeed, which will see big winners and losers in this race to dominate the AI world.
In the enterprise application space, the direct impact is founders expecting high valuations for anything “AI”. But even the successes mentioned above in “vibe coding” are significantly slowing down. Initial curiosity and time saving leads to high revenue, but can vanish when the expectations of end-to-end benefits are not met. With gross margins much lower at 25-50% (70-90% for SaaS) and at the mercy of the pricing of the infra layer the path to profitability is unclear. Enterprises are more than willing to experiment with AI, but they need reliably correct outcomes. A hard reality check was ”The GenAI Divide: State of AI in Business 2025,” from MIT, that reported 95% of AI projects at enterprises failed!
So, while most VC’s and PE’s are jumping on to the AI-First bandwagon, we at Pentathlon believe that AI will eventually run the enterprise world, but not before the “bubble” bursts. We simply see AI following the time-tested path of all Technologies – the Hype Curve!
Pentathlon way out
But who do we think would win in the steady state where AI rules? It is very interesting that in spite of the huge advances AI has made in the last 2 years, our thinking is an extension of what we have said earlier. Two years back, our thesis for Fund2 was use-case first, rather than Tech first. Last year we outlined how our early stage B2B Tech startups can provide better AI solutions than the Tech giants for niche enterprise use cases, using enterprise data with strong domain and industry knowledge.
While enterprises want to try out AI, they prefer evolution to a revolution. Hence the niche use-cases are better targets at enterprises than big-bang transformation of the entire industry. These champions of their niches (or SLM’s – “serviceable leadable markets”) will have a path to market leadership through adjacent spaces. And if they are building AI solutions profitably, they will be backed by VC’s and PE’s leading to IPO’s, or acquired by strategics.
But for a startup to become a champion, incremental value-add in activities is not sufficient. Everybody will achieve that with AI. The champions will stand out by using AI to capture the expertise and wisdom of the elite, whose decision-making makes the real difference. With AI, they will create very different value for their customers than the incumbents. Attribution of that value, will get them their fair share with pricing based on outcomes.
For that, the engagement with the enterprises will not be just the “arms-dealing” of AI products but the foot-soldiers providing the services to realise those outcomes for enterprises. SIs are seeing traction at enterprises for their pick and shovel offerings of building AI pipelines, data cleaning and migrations. But AI startups collaborating with SIs, will bring in disruptive value to the enterprises. Human agents working hand-in-hand with AI agents will bring in outcomes reliably. And combining appropriate Tech with AI will bring in efficiency in solutions leading to a profitable business, without requiring humongous funding. This requires deep industry knowledge. Not just an AI-First approach, that sometimes seems like a hammer looking for nails.
For Indian startups, enterprise solutions and developer tools remain the sweet spot of global opportunity. We see these opportunities growing beyond the traditional North American markets to new markets such as the Middle East. And indeed the non-metro Indian market as well, where the non-English speaking population, largely excluded so far in verticals such as Financial services, Healthcare, Education / skilling and Agriculture can be served with AI, hugely facilitated by our multi-lingual and multi-cultural data as well as our Digital Public Goods and Infrastructure (DPI/G). It is our hope that AI will help smart entrepreneurs overcome the traditional challenges of Tech adoption and pricing in India.
This will require going beyond conventional Go-to-Market strategies for B2B enterprise sale with product-led growth (PLG) for small enterprises (B2b), community-led growth for developer tools (B2C2B) and intermediary-led growth for B2B2C businesses such as banks, hospitals, schools and colleges. The intermediaries such as banking correspondents, nurses, teachers and village level entrepreneurs will be hugely empowered by AI and in turn, will ensure successful outcomes for end-beneficiaries, even better than the end-beneficiaries using AI themselves.
In conclusion
In my next articles I hope to elaborate our thinking on the aspects outlined above such as the verticals and use cases we like, their GTM’s, customer engagement models, pricing and even possible exits. And then AI itself keeps on evolving at breakneck speed! Exciting times. Stay tuned!
