Nebius raised five billion dollars, and the structure of the raise teaches more than the sum. It also explains something broader: why growth companies are compelled to keep raising.
Errors or inaccuracies are possible. Spotted something that looks wrong? Write to me and I will correct it.
What Was Raised
The figures here come from the company's own pricing announcement, dated 19 August 2026.
| 2030 series | 2034 series | |
|---|---|---|
| Size | $3.0 billion | $2.0 billion |
| Coupon | 0.50% | 4.50% |
| Maturity | 15 February 2030 | 15 February 2034 |
| Conversion price | $313.46 | $324.65 |
| Premium to market | 40.0% | 45.0% |
The offering was upsized from the $4.5 billion announced the day before. Net proceeds, after discounts and expenses, are estimated at about $4.94 billion. Settlement is expected on 24 August.
The reference price for both series is $223.90 - the stock's close on 19 August, the day it fell 9.87%.
And the Mechanism That Changes the Whole Picture
Both series carry an accretion mechanism on the principal.
And this is the part that vanished from almost every report on the deal.
In an ordinary bond, whoever paid 100 gets 100 back at maturity, and collects a coupon along the way. That is not the case here.
The 2030 series redeems at 110% of principal. The 2034 series redeems at 125%.
That is, the principal itself grows over the life of the note, according to an accretion schedule set out in the indenture. Whoever paid $1,000 will receive $1,100 or $1,250 at maturity - on top of the coupon.
So a "0.50% coupon" is not the cost of the financing. It is only the part paid in cash twice a year. The second part accrues quietly and is paid at the end.
And here is how that looks once you compute the effective yield to maturity:
| Stated coupon | Redeems at | Years | Effective yield | |
|---|---|---|---|---|
| 2030 series | 0.50% | 110% | 3.48 | about 3.17% |
| 2034 series | 4.50% | 125% | 7.48 | about 6.89% |
So on the first series, the true cost of financing is more than six times the number in the headline.
That calculation is mine, not the company's - but its direction is confirmed in the pricing announcement itself, in a way worth quoting.
Nebius itself discloses two different conversion prices.
The stated conversion price is $313.46 and $324.65, premiums of 40.0% and 45.0% over the market price.
And the effective conversion price at maturity, once the accreted principal is taken into account, is $344.81 and $405.82 - premiums of 54.0% and 81.3%.
Two pairs of numbers for the same deal. The first is what the indenture says; the second is what the investor actually pays.
And the difference is not technical. A premium of 81.3% means a holder of the 2034 series profits from conversion only if the stock nearly doubles. Until then, what they hold is a bond yielding about 6.9%.
A technical point worth knowing: conversion itself is computed on the original principal, not the accreted principal. So the accretion makes the debt more expensive without increasing the number of shares created on conversion.
And the Second Half of the Deal: An Exchange Already Executed
Alongside the pricing, Nebius entered into exchange agreements with holders of two older series - notes carrying a 2.00% coupon due 2029 and a 3.00% coupon due 2031.
It is exchanging $400 million of each - $800 million in total - for about 15.8 million shares.
And this number has to be read correctly, because it is easy to get wrong.
15.8 million shares is about 7.2% of the existing share count, and at market price they are worth about $3.54 billion.
It is tempting to conclude that the company gave up $3.54 billion for $800 million of debt. That reading is wrong.
The older series are convertible notes issued when the stock was far lower - its 52-week range starts at $63.26. Their conversion price is low, so they are deep in the money. Their holders already effectively held the right to those shares.
What the exchange does is not create new dilution - it turns dilution that already existed in potential into dilution in fact, and earlier. In return, the company removes $800 million of debt from the balance sheet.
And it still matters to the investor: shares created today are supply entering the market today.
The Dilution, Summed
| Shares | Share | |
|---|---|---|
| Immediate, from the exchange | about 15.8 million | 7.17% |
| Potential, from the new series | 15.73 million | 7.14% |
| Together | about 31.5 million | about 14.3% |
Against 220.4 million existing shares.
And Why Growth Companies Are Compelled to Raise - the Explanation
This is the broader question the deal illustrates, and it reaches far beyond Nebius.
The structural problem: the spending precedes the revenue by years.
A company building compute infrastructure pays for servers, for GPUs, for power and for buildings - before there is a single customer paying for them. A data center is built years before it is full.
And Nebius's revenue over the trailing twelve months is $1.36 billion. The current raise, $5 billion, is 3.7 times larger than that.
And a gap like that cannot be funded from operations, because the company's operating margin sits around zero. There is no profit to reinvest.
So three routes remain, and each has a price:
Issuing shares - dilutes immediately, and at today's price. A company that believes its stock is worth more tomorrow experiences that as selling cheap.
Ordinary debt - does not dilute, but demands a high interest rate from an unprofitable company, and usually collateral as well.
Convertible notes - a low cash coupon, and dilution that is deferred and conditional. The investor gives up interest in exchange for the option to become a shareholder if the stock rises enough.
And that is why this instrument dominates the sector: it lets a company sell shares at 40% above market, and only if the market gets there.
And what this deal adds to the explanation is that the price is not zero. The accretion on the principal is precisely where the cost returns: the company pays little cash today and undertakes to return more principal at the end.
And Why the Stock Fell Almost 10%
On the day of the announcement Nebius shares fell 9.87% and closed at $223.90.
This is a familiar reaction to convertible issuance, and it has two components.
The first is dilution - a widening of the share base, immediate from the exchange and potential from conversion.
And the second is technical, and it is usually the larger of the two. A substantial share of convertible buyers are funds that are not betting on the direction of the stock: they buy the note and short the stock at the same time, in order to isolate the option component. That short selling happens precisely on issuance day, and it is real selling pressure.
And the announcement itself says so explicitly - it notes that holders of the older series may sell the shares they receive and run hedging positions, and that such activity could depress the share price.
So a fall on the day a convertible is issued is not necessarily the market's judgement on the deal. Part of it is mechanics.
And the Pattern: $12.5 Billion in Under a Year
| Date | Size | Instrument |
|---|---|---|
| September 2025 | $2.75 billion | Convertible notes |
| March 2026 | $4.0 billion | Convertible notes |
| July 2026 | $0.775 billion | Secured debt - the company's first |
| August 2026 | $5.0 billion | Convertible notes |
About $12.5 billion in total over eleven months - 9.2 times the entire annual revenue.
And each round is larger than the last.
הזווית שלי
דעה אישית של אילן אברמוב - לא ייעוץ ולא המלצה
What I take from this deal is mainly an exercise in reading.
"A $5 billion raise at a half-percent coupon" sounds like almost free money. It is not. The effective yield is about 3.17%, and the company itself discloses that the real conversion premium is 54% rather than 40%. Both numbers sit in the same announcement - one in the headline and one in the body.
And that is a pattern that recurs in every complex financing instrument: the number that gets quoted is the one that is convenient to quote. What decides the matter is in the clause explaining how the principal behaves.
And what I think is worth holding about the structure itself is that there is nothing deceptive in it. A convertible note is a sensible trade for both sides: the company gets cheap cash today and pays in dilution only if it succeeds, and the investor gets the protection of debt with an option on the upside. The problem is not the instrument - it is how it is read.
And what I will follow at Nebius is not the next raise but when the business starts to fund itself. Four rounds in eleven months, each larger than the last, on revenue of $1.36 billion - that is a pace that describes a company in a building phase, not an operating one.
That is entirely legitimate at this stage, and it is also what defines the risk: a company that depends on the capital markets to continue building also depends on those markets staying open. When they are open, the structure works. The question is always what happens the time they are not.
(It is important to stress: this is my personal opinion only, and nothing herein constitutes a recommendation to take any action.)
