Cloudflare takes on AWS

I’m part of a small SaaS outfit where I’m the finance guy. I’m not a techie. But our head of engineering is, and he, who never really gets excited about anything, alerted the Exco today that he believes there is disruption coming to AWS egress pricing.

This is important because we have worked together for years and this is the first time he posted anything about our infrastructure costs (we’re an AWS and Google shop) to the Exco.

So I had a look at what he was excited about.

He posted two links. The first is the blog post dated yesterday outlining what Cloudflare are doing in the storage space. However it’s not yet generally available, you can sign up to a waiting list for access. The heading is "Announcing Cloudflare R2 Storage: Rapid and Reliable Object Storage, minus the egress fees”. I recommend reading the whole thing; below are some snippets:

“We’re excited to announce Cloudflare R2 Storage! By giving developers the ability to store large amounts of unstructured data, we’re expanding what’s possible with Cloudflare while slashing the egress bandwidth fees associated with typical cloud storage services to zero. Cloudflare R2 Storage includes full S3 API compatibility, working with existing tools and applications as built.”

→ So R2 is built to take on AWS specifically. The tools included are specifically targeted at S3, AWS’s product, and intended to make it seamless to switch to the customer.


“Egress bandwidth is often the largest charge for developers utilizing object storage and is also the hardest charge to predict. Eliminating it is a huge win for open-access to data stored in the cloud.”

→ So they are targeting the biggest piece of revenue to AWS (biggest cost to devs) for object storage.

“Our object storage will be extremely inexpensive for infrequent access and yet capable of and cheaper than major incumbent providers at scale.”

→ They will compete on price. They are making the same thing much, much cheaper and offering the same, or better, functionality.

The second link is a comparative calc of AWS pricing vs the new Cloudflare pricing:…

→ Here the bottom line from the thread, and without having redone the calcs - is that something that would typically have cost $59,247 on AWS will now cost $0.13 on Cloudflare.

I did a double-take when I read that. Not half price, not 80% off, no. They basically take something that has a significant price tag on AWS and then all but give it away for free.

I like this way of doing things about Cloudflare as a consumer. They are truly innovative and put out a lot of new products and services which techies seem to be increasingly breathless about.

However I dislike that they are not increasing either their top line or their bottom line to the same extent as an investor. Which is why I sold out completely a couple of months back.

This move exemplifies this dynamic for me. They are not capturing the $59k which is currently going to AWS, they are replacing that $59k with $0.13c. So in stead of capturing some of the value they are passing all of that value directly to the consumer. And in doing so they are crossing a pretty important line by openly crossing swords with AWS.

Perhaps I’m overblowing the extent or the relevance of this new price competition, or perhaps it is not as significant to AWS, or perhaps it does not have as big an impact or uptake.

But regardless of how this plays out, this is exactly why I love the company Cloudflare, but do not like the investment Cloudflare.

They are not, to my mind at least, running this company with a focus on being a profitable machine. They are making losses head over fist, and seem quite content to continue doing so. If there is value to be captured, they would rather pass most of it on to consumers than capture too much of it for themselves and us shareholders.

Perhaps others who are closer to this can also weigh in here, but taking the calcs in the twitter feed on face value means going on a price war with AWS. And however that’s dressed up just does not seem like a smart move to me.


(No position in NET, but it’s on my watchlist)


Thanks, WSM

That Twitter thread is great at explaining things from a developer’s point of view.

Before we get too worked up about Cloudflare dropping the economic ball, let’s remember that in the same way that Amazon set up S3 to support AWS in general, which they got into because they needed to have that capacity anyway for their underlying retail business……Cloudflare’s core business is serving up content on the web, for which they need to be storing content as part of the offering.

So I see this as a direct strategic copy of Amazon: take something that you have to be doing anyway and find a way to leverage that capability with new customers.




I think the financial comparison in the twitter thread is too exaggerated just to try to prove a point. 1 GB is a lot of data. As a comparison, this page is 300 kB on the first load, and on refreshes it is only 30 kB due to caching. is 3.6 MB on the first load and 146 kB on refreshes. Images are the source of the vast majority of page size.

Let’s get crazy and say your website has a 10 MB page size. In this scenario, 1 million unique page views would end up costing you $592 (1 GB = 1000 MB and 1 MB = 1000 kB). For revisits, the ratios for the mentioned pages were 10% or less. So 1 million non-unique page views would be about $59.

In the case of a small startup company, they probably see a lot fewer page views than 1 million, so they will end up paying less, such that these costs are negligible. Likewise for a larger company, their revenues should be large enough such that these costs don’t matter either. There is a point in company scaling at which these costs can become large enough such that they are worth reducing/optimizing, but I don’t see the cost as being a determining factor in the comparison between Cloudflare R2 or AWS S3 here.

There are probably use cases out there where 1 GB files are stored in AWS S3 and egressed, but I don’t know what they are off the top of my head. Maybe video streaming, but I think they use other technologies for storing data that are much more performant than S3.

As a developer, I am much more interested in the performance numbers, which were not provided in the blog post. My guess is that Cloudflare would be faster because of where they are storing the data, but we will have to wait and see.

Also, the design makes it easy to transition from S3 to R2, and I think that is pretty huge. If R2 is cheaper, faster, and easy to transition to, then it will gain market share. Then, Cloudflare can increase prices.

long NET


“Your margin is my opportunity” - Jeff Bezos

Hmm, looks like prince/Cloudflare is using the Amazon playbook here. Weren’t people complaining about Amazon never making a profit a decade ago? This is just one of many developments that will be used to continue to funnel huge amount of customers into the Cloudflare Ecosystem and Workers platform. Amazon has to be making ridiculous profit margins on egress fees.

I can tell you Cloudflare isn’t doing this for charity (seems to be becoming a popular opinion now). The company is being lead by 2 (formerly 3, RIP Lee Holloway) extremely driven/genius founders. It’s usually the case where companies grow into the size of their /founder leadership team’s ability. I don’t think it will be any different here. The market is pricing in the potential for eventually a trillion $+ market cap. That’s why the valuation that everyone keeps complaining about/not understanding why it’s so “high”. It will always be relatively high to whatever current multiples SaaS are trading at, save for some huge technology shift. But Cloudflare is built with disruption at it’s core and has methods to make sure it keeps up with technology changes.

Long NET (16%), fwiw


cost $59,247 on AWS will now cost $0.13 on Cloudflare

This is not possible. Repeat, not possible. This seems like wishful thinking post.
I am not sure where these calcs came from but be suspicious. AWS S3 is super cheap.
In general, storage is cheap. Compute is expensive.

If these numbers are true, Cloudflare is in trouble. You can’t make money if you give the product for free.

If these numbers are false, your investment is based on a false thesis.

If Cloudflare is able to offer GPU for cheap undercutting AWS/GCP/Azure, then it would make more sense.


I think the point is not the cost of storage (R2 is slightly cheaper than S3), but the bandwidth charge for moving that data out of storage to an endpoint user.

AWS S3 apparently charges a significant fee of 9 cents per GB for data to move from their active S3 storage onto your web page. So, if you are running a business in which a LOT of people are moving a LOT of data through your web page, then AWS would be an expensive solution.

Cloudflare’s approach is based upon “edge computing” and “edge storage”; that is, the data and compute power are not centrally located in one of the relatively few AWS locations, but are instead using Cloudflare’s distributed nodes around the globe to be physically (and logically) closer to users. Since NET’s R2 is already an “edge” service, if you moved your 1GB of data from S3 to R2 once (for 9 cents), from that point there is no specified bandwidth charge when someone pulls that 1GB from R2.

It is not as if the cost of bandwidth suddenly disappeared. Just because it is not a direct charge to NET customers doesn’t mean it is free. Instead, that cost would be embedded in their subscription fees. I assume that if someone is using NET to distribute 1GB files to 1 million people, then their subscription fee would be significant. Focusing on ‘line item’ charges as opposed to total cost is a bit disingenuous.

Tiptree, Fool One guide and Market Pass home Fool, no position in AMZN or NET


Cloudflare’s approach is based upon “edge computing” and “edge storage”;

The concept of edge compute and edge storage is provided by all cloud providers.
There is nothing new Cloudflare is providing or offering.

Google has its own underwater optical cable infrastructure around the world.

Unless Cloudflare has found a new innovative way to optimize its own storage and transportation costs, reducing price for customers just puts added pressure on margins.

1 Like


The part Cloudflare is offering at 0 cost is the transfer of data out of storage, not the storage itself. This is what egress fees are. It’s almost pure profit for AWS because there is negligible costs associated with it and is just paid because the data is “locked in” to the network.



Ben Thompson has a pretty good breakdown on this news here


Here’s an excerpt from Ben Thompson’s post at Stratechery explaining the basics of Cloudflare’s opportunity with R2 (just as Bnh references above):

"The reason that Cloudflare can pull this off is the same reason why S3’s margins are so extraordinary: bandwidth is a fixed cost, not a marginal one. To take the most simplified example possible, if I were to have two computers connected by a cable, the cost of bandwidth is however much I paid for the cable; once connected I can transmit as much data I would like for free — in either direction.

That’s not quite right, of course: I am constrained by the capacity of the cable; to support more data transfer I would have to install a higher capacity cable, or more of them. What, though, if I already had built a worldwide network of cables for my initial core business of protecting websites from distributed denial-of-service attacks and offering a content delivery network, the value of which was such that ISPs everywhere gave me space in their facilities to place my servers? Well, then I would have massive amounts of bandwidth already in place, the use of which has zero marginal costs, and oh-by-the-way locations close to end users to stick a whole bunch of hard drives.

In other words, I would be Cloudflare: I would charge marginal rates for my actual marginal costs (storage, and some as-yet-undetermined-but-promised-to-be-lower-than-S3 rate for operations), and give away my zero marginal cost product for free. S3’s margin is R2’s opportunity"


I don’t understand why AWS S3 and egress of data from it is so important.
Just because Cloudflare CEO talks about it does not make it relevant.

It is like focusing on the cost of a door handle of a car.

It is a very small part of what enterprises look for a cloud providers like AWS, GCP, Azure offer.
The broader cloud eco system and capabilities is much more important.


It is like focusing on the cost of a door handle of a car.

It is a very small part of what enterprises look for a cloud providers like AWS, GCP, Azure offer.

Data transfer fees can cost companies north of $1,000 PER transfer when transferring ~10TBs of data (which is not that much data anymore).

A much better analogy would be focusing on the fuel efficiency of a car you are buying, not the door handles. Cloudflare is the Tesla with infinite MPG (no data transfer fees). AWS/other cloud infra providers are similar to a 2013 Hummer @ 10mpg. Prince has stated a few years ago that he thinks margins for egress @AWS are over 99%. They are definatly a large hidden cost for alot of companies, so I am not sure why you are claiming they aren’t.

The broader cloud eco system and capabilities is much more important.

Agreed, this is what Cloudflare is building and constantly releasing new features on. We will see how it goes, but disruption seems to be happening.

Good luck


For those of you interested, I found a very nice explanation of how Cloudflare can pull off its R2 offer:

Basically, the explanation runs as follows: AWS charges egress fees (but not ingress), so that data is effectively locked in, after a costless buildup phase (since ingress = 0). These egress fees are a big source of profit and margin to AWS.

Cloudflare doesn’t have to (and doesn’t want to) charge these fees, because the bandwidth is already there and paid for. Cloudflare chooses to attract new clients by this free ingress-egress offering, and it doesn’t cost them a dime! (Clients will have to pay for storage though.)

So it’s a win-win: Cloudflare gets new clients for free by using installed capacity and these clients pay a lot less than for AWS. Bonus for the clients: they’re not locked in anymore, since they can exit at will. Your margin is my opportunity.

I suppose that for Cloudflare, the goal is to upsell other services to these clients. Whether this’ll work remains to be seen, but we cannot call this move stupid.


Cloudflare doesn’t have to (and doesn’t want to) charge these fees, because the bandwidth is already there and paid for. Cloudflare chooses to attract new clients by this free ingress-egress offering, and it doesn’t cost them a dime! (Clients will have to pay for storage though.)

I think this stretches the notion of ‘your margin is my opportunity’ as this isn’t quite a race to the bottom (of pricing) but rather straight to the bottom. Wouldn’t NET’s competing egress offering have been just as appealing at 50% discount rather than 100% discount. Saul’s concerns with Cloudflare’s altruistic vs commercial priorities ring louder in my head now that I can see a clear example of a missed revenue opportunity