Issue 05 — 10,000 one-person conglomerates by 2028, and the metric that tells you if you're really one

Sunday evening. I sat down with a blank spreadsheet and four columns: venture, monthly P&L contribution, my hours per week on it, and one derived column — ₹ per operator hour.

I filled in V4. Then V5. Then two more ventures I do not name publicly yet. The first two numbers were flattering. The third was fine. The fourth was embarrassing.

One of the four was not a business. It was a hobby I was calling a business because the word “portfolio” sounds better than “hobbies.” A ₹/hour of ₹780 is not a venture. It is a Sunday pastime with a Stripe account.

I am writing this issue because I think 90% of the people who will claim to be one-person conglomerates in 2028 will have at least one row on that spreadsheet that looks like my fourth row, and will not have made themselves look at it.

PRED-005 — By December 2028, at least 10,000 solo operators will run five or more revenue-generating businesses simultaneously, with AI agents handling 60%+ of operational execution across all ventures. Not passive side projects. Active businesses with customers and revenue.

Confidence: 4 out of 5.

The model works. I run it. The author of this book runs it. That is not marketing — it is the practitioner receipt the chapter is built on. V4 and V5 — agent security scanning and vibe code quality analysis — were built in parallel by one person with agents, delivering capabilities comparable to multi-person engineering teams. Agents handle roughly 80% of execution. Humans decide and orchestrate. Agents execute. That three-word cadence is the whole operating model, and the reason the model scales to five ventures instead of one and a half.

The surrounding evidence is not subtle either. Daniel Vassallo popularised the portfolio-of-small-bets thesis in 2021 and runs five-plus revenue streams himself. IndieHackers has more than 100,000 members, with increasing reports of operators running three to five projects. The #buildinpublic community on Twitter shows thousands of solo operators sharing multi-project portfolios. Inference costs dropped roughly 10x between 2023 and 2025. The trajectory is clear. The question is only how fast the curve bends.

The chapter makes that case. What it does not do — because it is a chapter, not a newsletter — is give you a diagnostic for whether you are actually running five businesses or five hobbies. And that is where the 10,000 number falls apart in practice. Ten thousand people claiming to run five-venture portfolios is easy. Ten thousand whose portfolios survive a financial audit is a different number.

So here is the diagnostic I run on my own portfolio every quarter. Call it P&L per venture per operator hour. One formula:

(monthly P&L contribution from the venture) ÷ (your hours per week on the venture × 4.33)

That is it. No adjustments. No “strategic value.” No “it’ll pay off later.” Just the ratio.

Pick a threshold and hold yourself to it. Mine is ₹5,000 per operator hour (roughly $60). Below that, the venture is a hobby dressed in business language. I used to resist this number. Every founder I tell it to resists it. The resistance is the signal — the number is always a venture you wanted to believe in.

Here are the rough receipts from the spreadsheet I filled in on Sunday, rounded to the nearest five hundred so I am not leaking commercial detail: V4 runs comfortably above the threshold. V5 clears it. A third venture sits at roughly 1.6x the floor — fine, not flattering. The fourth clocked ₹780 per operator hour. That venture is either getting agent-ised harder in Q2 or getting killed in Q3. No third option.

This is the number that separates PRED-005 from wishful thinking. If the 10,000 operators forecast for 2028 each have five ventures, that is 50,000 venture-rows on 50,000 spreadsheets. If the median of those rows cannot clear a ₹3,000/hour floor, the prediction is directionally right and substantively empty — the 10,000 number came from counting anyone with five landing pages. If the median clears the floor comfortably, I am looking at a new business category.

I want the new business category. The only way to get it is for the people running portfolios to measure the embarrassing row first.

The published falsification trigger:

If by December 2028, the one-person conglomerate model remains niche with fewer than 1,000 identifiable practitioners, or if no major business publication covers it as a trend, this prediction is wrong.

That is the stake as written. Here is the way I am actually most likely to lose it.

Not headcount. Economics. A 2028 world with 15,000 self-described one-person conglomerates where the median venture row sits under ₹2,500 per operator hour is the prediction being technically right and commercially fake. The category exists but the unit economics do not. That is the failure mode I take seriously, and it is the reason I publish the ₹/hour metric alongside the headcount number. Counting people who claim the title is easy. Counting ventures that clear the floor is the real falsification.

If you run three or more ventures — not side projects, revenue-generating — I want your ratio table.

Reply to this email with your venture count, your rough monthly P&L contribution per venture (anonymised on request), and your honest hours per week per venture. I will compute your ₹/hour distribution and publish the aggregate — no names — at atin-agarwal.com/predictions/pred-005-solo-operator-conglomerates/, with credit if you want it. Quarterly updates as data comes in.

What I am actually watching: the median row across operators running five or more ventures. If that median cannot clear ₹3,000 per operator hour by 2028, PRED-005 is structurally fragile regardless of what the headcount number says. That is the distribution I will publish, and the distribution I will hold myself to.

If you are a solo operator running three or more ventures: compute the number this weekend. Do not compute it in your head. Open the spreadsheet. Four columns. You will find one row that you have been protecting from the math. That row is the work.

If you are a founder running one venture and thinking about adding a second: know the floor your existing venture hits per operator hour before you add anything. If venture one is at ₹4,000/hour, adding venture two pushes it down before it pushes it up. The second venture has to clear a higher bar than the first, not a lower one.

If you are an investor looking at one-person conglomerate pitches: P&L per operator hour is the unit economic that lets you evaluate a one-person portfolio the way you would evaluate a fund. Ask for the ratio table in the data room. A founder who cannot produce it on demand is telling you the answer.

If you are an operator inside a larger company watching this category from outside: the interesting question is not whether you could run five ventures. It is whether your current day job would survive its own ₹/hour audit if you ran it on yourself.

This issue is drawn from Chapter 9 of The AI Agent Economy — 15 falsifiable predictions with dates, numbers, and explicit triggers for being proven wrong. Pre-order on Kindle — $9.99. Release July 1, 2026. atin-agarwal.com/books

Read the full PRED-005 entry on the public tracking page → atin-agarwal.com/predictions/pred-005-solo-operator-conglomerates/