Jay Kalam

Built deals worth $500M+. Built data products for 10 years. Now building AI tools with zero engineering background.
Frameworks, lessons, honest takes.

[Jay Kalam @jay_kalam]

I spent years in M&A. Sat in deal rooms, built financial models, ran due diligence on transactions worth hundreds of millions.

Now I build AI products. And the thing that strikes me most about the AI startup market is how few of these companies will actually get acquired in the traditional sense.

Most will be acqui-hired. The company dies. The team gets absorbed. The product gets shelved.

Here's why.

SaaS M&A has a formula. You're buying recurring revenue. Predictable cash flows. Customers with switching costs. The entire valuation model is built on "what will these customers pay next year?"

AI startups don't have that. Most have no moat, no lock-in, and an LLM provider that can replicate their core feature in a sprint. The wrapper thesis is real, and most AI startups are wrappers.

So what do AI startups actually have?

People. Engineers who've shipped production AI systems. Researchers who understand fine-tuning, evals, and inference costs at scale.

That's what the big acquirers are buying. Not the product. Not the IP. The team. The "acquisition" is really a hiring event with a retention package.

This accelerates in 2026-27:

1. Foundation model costs keep dropping. The wrapper moat gets thinner by the month
2. Big tech has unlimited capital but a genuine talent bottleneck
3. Most AI startups will hit a growth ceiling once the novelty wears off and enterprise buyers start doing proper due diligence

The math favours acqui-hires over full acquisitions.

What this means if you're building:

- Your team IS the asset. Invest in people, not features nobody asked for
- Don't raise on inflated valuations. It makes acqui-hire math impossible for buyers
- Build genuine integrations with enterprise customers. That's one of the few things that creates real switching cost
- Document your thinking, not just your code. Acquirers want to know HOW your team makes decisions

The counterpoint: some AI companies WILL build real moats. Proprietary data. Domain-specific models that foundation providers can't easily replicate. Network effects from user-generated training data.

But these are exceptions. If you can't articulate your moat in one sentence, you're probably building an acqui-hire target. That's not necessarily bad, but you should know which game you're playing.

I'm writing more about where finance meets AI. The stuff the deal room teaches you that the tech world doesn't talk about.

What's your moat?

#MergersAndAcquisitions #AIstartups

[Jay Kalam @jay_kalam]

World Cup begins, and so will the England excuses..
#SomosMéxico
@ITVSport @BBCSport

[Jay Kalam @jay_kalam]

@sairahul1 Are you still cold?

[Jay Kalam @jay_kalam]

@PeterDiamandis Are you just paraphrasing Hemingway?

[Jay Kalam @jay_kalam]

@gregisenberg shared Sequoia's "services is the new software" map. Over $1T in professional services they reckon AI agents will replace.

I spent a decade inside two of those boxes. Executive search and management consulting. Both in what Sequoia calls "Copilot Territory." I've sat in the rooms where those fees get justified, and the map gets something important wrong.

It treats all services as if the value is in the work. It's not. In consulting and exec search, the client isn't paying for the analysis or the candidate shortlist. They're paying because a CEO will take the partner's call at 7am on a Sunday. That's relationship arbitrage, and no agent replicates it.

The intelligence-heavy boxes - payroll, KYC, claims adjusting - those are pattern matching at scale. AI genuinely eats that. But the judgement-heavy ones? They get more profitable, not less. I watched it happen in market intelligence. The research got automated. The price didn't drop. Margins just widened for the firms that moved first. The fee stays because you're buying judgement and access, not hours.

Sequoia's looking at this through a VC lens: where can software eat services? From inside the industry, the question's different. It's whether existing firms adopt fast enough to keep the margin, or whether AI-native entrants undercut them. In exec search, most firms still run on spreadsheets and instinct. In consulting, deliverables still get built by analysts pulling 80-hour weeks. The distance between what AI can do today and what these firms actually use is enormous.

That gap is where the real money is. Not replacing the service. Delivering it at a fraction of the cost while the invoice stays the same.

Everyone can see the river. Very few know where it's crossable.

#consulting #executivesearch #AIstartups

[Jay Kalam @jay_kalam]

Just published my first long-form piece on X.

What M&A due diligence taught me about evaluating AI startups. 4 frameworks from the deal room that most founders have never thought about.

If you're building, investing in, or evaluating AI companies, this is for you.

#MergersAndAcquisitions #AIstartups

[Jay Kalam @jay_kalam]

https://t.co/dJ4WcSBfoS

[Jay Kalam @jay_kalam]

This is basically a PE roll-up strategy automated at zero marginal cost. In private equity you acquire 10 companies in a niche, kill the ones that don't work, double down on the winners. You're doing the same thing but with ideas instead of companies, and validation costs almost nothing.

The interesting question is what happens when everyone can do this. In PE, the edge moved from "finding deals" to "operational improvement after acquisition." Same thing will happen here.

[Jay Kalam @jay_kalam]

Loving @AIatMeta superintelligence. Think they need to spend a few hundred billion more
#llama #LLMs

[Jay Kalam @jay_kalam]

@boringmarketer @gregisenberg 100%

[Jay Kalam @jay_kalam]

@gregisenberg Check out the invest like the best podcast episode with gavin baker and you’ll have your answer. Clue: the company that’s first in line to get the most of NVIDIAs next gen gpu
@InvestLikeBest

[Jay Kalam @jay_kalam]

@rileybrown Whatever he’s paying, I’ll pay you 101% more.

[Jay Kalam @jay_kalam]

@tim_cook Is the only thing that has changed, everything?

[Jay Kalam @jay_kalam]

@BloombergTV @FerroTV @lisaabramowicz1
Just a minute if your time on Bloomberg please!

[Jay Kalam @jay_kalam]

@thisiszanc @antonosika @lovable Access

[Jay Kalam @jay_kalam]

@ShaanVP 1. A stagnant mind is the only true career ceiling. Wake up curious, sleep upskilled.
2. Chase purpose, not titles. The former compounds; the latter fades.
3. Career growth isn’t just climbing ladders - it’s choosing which walls your ladders should lean against.

[Jay Kalam @jay_kalam]

Hey @sama, my 7-year-old loves chatting with ChatGPT voice while we’re in the car. I’d love to set up his own account under my profile with parental controls. Any update on when supervised accounts will be available?

[Jay Kalam @jay_kalam]

It is the beginning of every achievement, the conclusion of every saga. It sparks ambition in the heart of struggle, and seals the finale of every drama.

What is your guess?

#Riddle #Puzzle #Wordplay #Brainteaser

[Jay Kalam @jay_kalam]

@peterkay_co_uk have they reached out to you to play the lead for the #lukelittler biopic? Surely you’re a shoe-in to play The Nuke. #WorldDartsChampionship