Why Does a 0% Bond Cost So Much?

The $2.5bn Question

I am sat in a quaint little coffee shop in Hanoi, sipping my egg coffee, mindlessly scrolling through my phone when Alfie shows me an article about Cloudflare’s 0% $2.5bn bond deal. And I think to myself: what?

Cloudflare has raised a total of $2.5bn without paying its investors a single penny of interest. But much to my dismay – and perhaps yours too – there is no such thing as free money. The deal is a convertible bond, which means at some point, investors will have the option to convert their debt into Cloudflare shares at a fixed price.

I take another sip of my coffee and click on the link to the terms of the deal.

The notes are five-years and mature in August 2031. Investors paid par 13 Aug 2026 and receive $100 at maturity – unless Cloudflare’s share price rises sufficiently such that converting shares is worth more than the cash. The conversion price is $496.94 per share, representing a ~60% premium to Cloudflare’s $310.59 reference price when the deal was struck. Investors have effectively said: I will lend Cloudflare money for five years at 0%, but in exchange, I want the right to participate in a large equity upside. Which sounds reasonable, until you ask the more important question: how much is that option actually worth?

Deal Terms

So What Did Investors Actually Buy?

Cloudflare 2031 Convertible

• Issue Price: 100
• Cloudflare share price at pricing: $310.59
• Conversion: $496.94/share
• Conversion premium: 60%
• Capped call cost: $259.5mn
• Capped call price: $854.12/share
• Rating: Unrated
• Seniority: Senior unsecured
• Existing converts: $1.125bn 2026s + $2.0bn 2030s

A vanilla zero-coupon bond would normally need to be issued at a discount to par, so the investor’s return comes from the bond’s pull to 100 at maturity. Here, however, investors paid 100 despite receiving no coupon; they didn’t just buy a bond, investors bought the right to participate in Cloudflare’s potential share-price upside.

The convertible isn’t just a bet on Cloudflare’s share price, it is also a bet on Cloudflare’s credit worthiness, and this is where the valuation gets interesting:

Valuation

Breaking the Bond Apart

To determine whether this option was valuable, I built a credit valuation model. There are, however, a few caveats. It is unrated, and so I assign it a B spread note since it is high growth with high cash consumption, and the lack of an established credit rating implies a materially higher credit risk than its IG peers. This translates to a spread of 290bps and under a tighter BB case, a spread of 160bps, based on ICE BofA credit-spread data; 350bps is my judgemental downside case.

At a 290bp base credit spread, the bond floor is 69.4 per 100, and investors paid 100, meaning they paid 30.6 points for the embedded equity option. Across the full deal, that optionality is worth roughly $765m. On a per share basis, that is the equivalent to paying around $152 for a five-year $496.94 strike call versus the model value of roughly $135. So, will the option really be worth that much? I faced the same problem paying 65,000 dong for the coffee and croissant. The number means nothing until you realise it is £1.80. The difference here is after that conversion, I was pleased. Investors here shouldn’t be, they are paying (according to my analysis and assumptions) $152 for a call that is worth $135.

Now volatility comes into play. For an embedded option, investors want this because the higher the volatility of the underlying stock, the greater the value of the embedded call. Cloudflare, historically speaking, has been very volatile. Its realised volatility 1-year is around 62%, 3-year is 55%, and 5-year 69%. Given that the convertible bond is 5-year dated it may seem obvious to select an implied volatility of 69%. However, the previous years include the 2021/22 software sell-off, where Cloudflare faced an 80% decline in share price. On the other hand, 45% would be equally as aggressive, so I settled on 60% implied volatility. That consolidates the 3-year volatility and applies a 10-15 percentage point discount that convertible arbitrage investors sometimes use.

At 60% volatility, the model values the embedded option a value of 27.2 points per $100 of face value.

69.4 + 27.2 = 96.60

So, on my simplified framework, the notes are worth 96.6 when investors paid 100. In other words, even after giving Cloudflare the benefit of a high implied volatility assumption, the European model does not get me to par, and the notes, are 3.4 points rich.

Convertible valuation (par)

• Issue price: 100.0
• Premium paid: +3.4 pts

Rather than endlessly argue over whether 60% is the ‘right’ volatility assumption, I reverse the question: what volatility does the market require for the deal to break even?

The answer: 67%

Credit

The Credit Spread

At this point, my back hurts from sitting hunched over the small desk in my hotel room, and it is now 4pm, an excellent time for another cup of coffee. Now that I am in yet another coffee shop, I am ready to move on to the part of the analysis that is less straightforward: the credit spread.

If the spread is tightened from 290bps to 160bps (i.e. Cloudflare is treated as BB instead of B), the bond floor rises from 69.41 to 74.08, meaning the total value rises to 101.27 (assuming the same volatility, equity option, etc.). Suddenly, the same financial instrument becomes slightly cheap rather than slightly expensive, and it serves as a useful reminder that valuation isn’t really a single-variable bet. My view on both Cloudflare’s volatility and what sort of credit they deserve to be then, matters.

Peer comparison proved to be useful to contextualise my assumptions, CrowdStrike and Akamai are rated around the BBB/Baa2 area (IG), whilst Datadog and Snowflake are also from the unrated software universe. Another bite of my croissant may help: Cloudflare’s balance sheet is strong in absolute terms, but its debt-to-free-cash-flow is weaker than its IG peers such as CrowdStrike. Thus, it becomes difficult to argue that Cloudflare should automatically receive the same credit treatment as stronger software names purely because they operate in similar industries.

Structure

The Fine Print

And then, of course, there is the fine print. The notes contain a soft call after Aug 2029, contingent conversion provisions and other terms that make the embedded option less valuable than holding a vanilla European call. I also include a haircut of 1 percentage point to help account for the potential risks, whilst small, it acknowledges the limitations and simplicity of the model.

There is another figure in the transaction I find particularly revealing. Cloudflare spent $259.5m on a capped call transaction, with the cap sitting at $854.12 per share (a 175% premium). The capped call is designed to reduce dilution if shares rise substantially – so I excluded it. But economically, Cloudflare raised $2.5bn and spent $260m to hedge the equity optionality it sold. Conversely, I have raised $0bn and spent $0 hedging my own optionality.

Positioning

The Verdict

So where does that leave me? Currently, with an empty coffee cup. My model on the other hand, relies on a 60% five-year volatility, a 290bp B credit spread and a one-point haircut. On those assumptions, I arrive at a fair value of around 95.6 against a market price of 100. So, I would SELL / AVOID, but only narrowly. It is not because I dislike Cloudflare as a company. In fact, the opposite is probably true: the growth story is exactly what makes the equity option so interesting. But the convertible is asking investors to pay for a significant amount of that upside upfront, and at my assumed credit spread it requires roughly 67% volatility simply to break even.

If I believed Cloudflare deserved a BB-like credit treatment, the valuation also changes meaningfully. At 160bps spread, 60% implied volatility, the bond is valued at 100.3. So, the trade sits in an interesting zone where a relatively small change in either my credit or volatility assumptions can move the recommendation from sell, to hold, or even buy.

Perhaps this is my most important takeaway from this project. The market is effectively asking investors to believe two things at once: that Cloudflare is a strong enough credit for the bond floor to hold up, and that its equity will remain volatile enough for the conversion option justify its price. I am reasonably comfortable with the second assumption. I am less comfortable with the first. Put the two together, and I don't quite get to 100.

By now, my coffee is finished and there is not much left of the croissant either (I am equally as surprised as you are). The LinkedIn article that started this whole ordeal is still open in another tab, but I am no longer impressed by the 0% headline. The deal is clever, and for Cloudflare, undeniably attractive. $2.5bn raised without a conventional coupon with the dilution protection of a capped call. But the investor pays for both their willingness to underwrite Cloudflare, whilst simultaneously paying for the equity optionality that requires a very high level of sustained volatility to justify itself.

Perhaps, then, the better description of Cloudflare's 0% bond is not free financing, but expensive optionality disguised as cheap debt. At the right credit spread, or with enough volatility, the numbers can certainly work. But under my central assumptions, I would rather keep my 100.