Fastly Q120 recap & its potential

FSLY - Q120

As mentioned in my Cloudflare write up, I own both Fastly and Cloudflare now (more Fastly, at 13% to 9%). I have had a lot of success with Cloudflare thus far, but Fastly has had triple the gains while I sat only in its competitor. I missed out on that, but started paying much closer attention once rev growth was projected to be 15pp higher QoQ, and ramped up quickly.

Shout out to Smorgasbord for his good coverage of Fastly recently. And another shoutout to SSI blog. His deep technical dive into Fastly’s architecture was utterly fantastic. (More on those at the end.)

Shareholder letter:…
SSI recap:…
CC transcript:…
Bear buys:
Bert take:…

Revenue 62.9M +38.1%
Adj Gross Margin 57.6% +60bps
Adj Op Loss -6.0M (vs -7.0M)
… margin -9.5% (vs -15.3%) !! +580bps
Adj EPS -0.06 (vs -0.30)
Opex 47.6M +40.5%
CFFO -7.2M (vs -10.1M)
… margin -11.4% (vs -22.2%) !! +1080bps
FCF -18.8M (vs -14.9M)
… margin -30% (vs -32.7%) -1500bps seq
Custs 1837 +13.3%

  • Enterprise Custs 297 +22% (88% of total)
    ARPU 642K +5.8% seq
    $NER 133% +300bps !!
    $NRR 130% ($NER + churn)
  • first Q with new CEO
  • tailwinds from usage based pricing
  • network is designed to scale, no impacts to speed or performance
  • expect higher usage for rest of year, guided +53.6% (52-56%) growth next Q, for +1550bps seq !!!
  • expanding availability of Compute@Edge beta in 2H20
  • example use cases cropping up in beta: edge content encryption, custom identity auth systems at edge, custom global DLP (data loss & protection) system, advanced ML at edge, personalization of content
  • continue to expand and scale edge network, no supply chain impacts thus far
  • increased usage in custs at forefront of tech innovation, like Shopify, Khan Academy, Pinterest
  • in 55 markets, w/ 2 new markets in Kansas City & St Louis
  • expanded POPs in existing markets, now have
    88Tb/sec of global capacity [then later expanded it to 100Tb/sec]

Announced $50M fund for helping non-profits & open source:…

Announce support for new cutting-edge browser protocol (HTTP/3 + QUIC):…

Announce Developer Hub:…

  • new dev tools to build fast, scalable and secure modern apps on Fastly’s edge cloud platform
  • “Fastly Fiddle” testing sandbox, with ready-to-deploy code snippets
  • highly-searchable pool of tutorials, ref guides and docs
    … if you want to see Developer Hub:

Achieved 1T/sec in global capacity:…

  • 72 POPs in 55 markets w/in 26 countries
  • now has 100Tb/s throughput to edge, expanding network capacity 35% over last 5mo
  • acquired Tesuto virtual network emulation platform: “By emulating networks at scale, Tesuto’s technology can be used to create sandbox environments that simulate the entire Fastly network, providing a view into the potential impact of a deployment cycle before it is put into production. … this technology will improve network design, enable faster hardware and software deployments, and enhance overall efficiency.”

Fastly found to have Amazon and IMDb as customers:…

  • significant as Amazon isn’t using its own AWS CloudFront service
  • some CDNs are trying to grab a larger share of the OTT video traffic and gaming downloads; Fastly has mostly stayed away from that commoditized side of the CDN market

My stance: Notice the lack of !! up top? The only thing that stands out in the numbers alone is margin improvements. Revenue growth dropped to 38% from 44% prior Q. But regardless of that dip back and the lack of immediate glitter in the numbers, it seems clear from the guidance that their extreme focus on high-end enterprise customers w/ heavy needs is paying off huge during this swell in web traffic during the pandemic. Their usage-based pricing model is getting a significant boost, to the tune of +15pp (probably more) of rev growth QoQ. QUITE A TAILWIND, I’D SAY. Ultimately, Cloudflare is a ~50% grower guiding down (hurt by being fixed rate in this environment of spiking usage), while Fastly is a ~40% grower guiding way higher (helped by having usage-based pricing). I lowered my NET and bought FSLY heavily.

They focus on a higher-end customer, and ARPU continues to rise +5.8% seq. That $NER of 133% is showing that these high customers, once landed, only spend more. Fastly’s low gross margins will be helped going forward by higher margin add-ons like Compute@Edge. Op and cash flow margins are moving upwards while FCF margin looks pretty flat YoY --however, last Q FCF margin was -15%, and this Q it is back to -30%. It seems building out edge networks can be costly, and they seem busy – they had 11.6M in CapEx this Q alone vs 19.5M all of FY19. Adding a POP is initially expensive, but luckily, then has way lower costs to maintain & expand once in place. They are signaling more expansion in 2020 - so don’t expect FCF to improve much in the near-term.

FSLY is extremely popular all of a sudden. And that is for a reason. They are a best-of-breed solution in edge networks. They have a top notch technical crew that solves difficult problems in a VERY future-forward way. This means an EXTREME detail to speed and performance, and an EXTREME long-term view over their operational architecture. Their edge network uses software-defined networking on lower-cost commodity networking gear. This allows every single aspect of their product line to be programmable and scalable. Remember those building block APIs that TWLO and OKTA are exposing to make their platform into usable components? Think that x100 with Fastly. This isn’t just APIs to do some set of operational behaviors, this is about customers being able to control the very global network their services and content run on.

I have been agreeing strongly w/ Smorgasbord’s series of high quality posts on Fastly. He has been all over the technical points in fighting against the dismissal of Fastly as being just a CDN in a commodity market. I believe the naysayers are solely looking through the lens of delivering content, so not viewing the potential here correctly at all. That is the old Fastly – the capabilities they built to get here, and that they have succeeded at thus far. This current Fastly needs to be thought of as one giant programmable global network, and soon are extending that programmability into edge nodes themselves (edge compute & storage).

Here is a point Smorgasbord keeps making: Fastly’s focus is on programmable edge, not CDN/Streaming. One thing that is critical to understand with edge networks like Fastly and Cloudflare, is that “programmable edge” means programmable EDGE NETWORK (the switches and the global interconnections) plus programmable EDGE SERVER (compute at points of entry). Things can then built on this combination. CDN is now just a specific application on top of this edge cloud platform, that Fastly leveraged to be able to fund this architecture and get to where it is now. But that is not the focus of the business anymore. Another potential application is Secure Web Gateways (Zscaler ZIA) or Zero Trust (Zscaler ZPA), as Cloudflare is now delivering. And there are countless more use cases & product lines & verticals to come, to be discovered by Fastly itself, or by their customers, who are building their own solutions on top of their platform. The potential here is unknowable, but sure to be large. Please keep that mind.

What is abundantly clear right here and now is that more and more web traffic is being generated. Cloudflare is rising to a steady 50% growth for a reason. Fastly is jumping from 38% to 54% growth for a reason. Owning these companies right now is not an aspirational gamble on future technologies! These are companies executing well, rising quickly on the tailwinds of the effects of this pandemic (rising web traffic from stay-at-home & remote working), so are having a lot of success just with their EXISTING programmable edge network and CDN capabilities. Edge compute is still to come – and will MAKE REVENUE GROWTH RISE MORE FROM HERE.

The writeup by SSI blog was a fantastic read, and I link it below. But it’s highly technical, along with many of the technical discussions on this board that I link below. I’ll be boiling it all down very soon from an investment angle, in a follow up post next week on what Edge Networks really are. Stay tuned.

long FSLY, NET


I summarized much of this but here are some good reads if you want to go deeper.

SSI technical review:…

  • its clear Fastly takes a non-standard approach, typically a harder-to-take one, in the name of performance and customization (developer focused)
  • software approach using SDN switches instead of expensive routers
  • intentionally putting fewer but more powerful POPs in (more compute, ram, SSDs) for faster delivery times
  • wrote custom storage system to squeeze as much performance as possible
  • all aspects of CDN are programmable & performance-focused
  • using customized version of Varnish Config Lang (VCL), and are heavy committers to open-source project
  • customers are generally cutting edge & tech saavy: Shopify, Stripe, Spotify, Slack, Twitter, Pinterest, GitHub, Amazon, Microsoft, Wayfair, Vimeo, Reddit, NYT
  • are doubling TAM with app services & security features (17.5B to 35.4B)
  • Compute@Edge announced Nov 2019, now in beta
  • provides dev environ that provides fast, secure, distributed compute at high scale
  • potential use cases in ML, IoT, gaming
  • beta customers surprised them by also doing more security
  • startup times in Compute@Edge are massively faster than competitors (FSLY 35us, NET Workers 3-5ms, AWS Lambda 200ms), and smaller mem footprint (FSLY 5Kb?, NET Workers 3Mb, AWS Lambda 35Mb)
  • going live with it 2021, no revenue til then

… Smorgasbord’s recap:…
… Ibuildthings technical decoder ring:…

Smorgasbord reviews competitive landscape:…

Ronjob on NET vs FSLY:…

Smorgasbord notes from investor presentation:…

  • TAM is growing
    * focus is on programmable edge not CDN/Streaming
  • single network across ~2500 servers in 72 POPs
  • allows customers to switch between cloud providers, contain costs
  • major events have so much scale, will get covered by multiple edge providers
  • middle-tier products emerging to handle all aspects of customer’s content
  • in 55 markets now, ultimately expect 100 w/o China
  • excited about edge computing and security products

hmc recapping Beth Kindig’s views on FSLY:…
Smorgasbord response:…

Smorgasbord on end device vs edge vs core:…

  • Edge compute’s potential is doing things the edge device cannot do, and where low latency matters (sending back & forth to core servers is too high)

Smorgasbord on edge computing vs CDN:…

  • edge computing is NOT equal to CDN
  • CDN content being cached comes from origin servers
    * in edge computing, data primarily comes from the endpoint devices, not from origin servers
  • edge computing is not solved by cloud providers doing region expansion or “bringing origins closer to the edge”
  • AWS does have edge compute solutions in Lamda and Greengrass (IoT) products

Smorgasbord on ramping up Compute@Edge beta to GA schedule:…

  • tying in observability
  • can feed metrics into Datadog, Splunk or Elastic

Nilvest on FSLY vs NET:…
[good summary, except, Nilvest, you are completely discounting that Cloudflare is in this same space (having a programmable network w/ edge compute) - this is not an either/or]

Ibuildthings recaps the discussion on FSLY potential:…

Smorgasbord with thoughts on some bear cases:…

  • best tech may not win
  • one CDN isn’t going to win all traffic, customers want to spread out
  • isn’t the best CDN for everything, security isn’t as strong as NET
  • edge compute is in beta til 2021, no vision into how good it will be
  • competition will arise from cloud providers

Nilvest on FSLY’s strengths:…

  • this is not vanilla CDN, it is an edge service platform with high degree of programmability [and scale]
  • they aren’t going after the commoditized CDN markets
  • picks and shovels play, for trends of e-commerce, work-from-home, online gaming, 5G and digital transformation

WillO recaps SSI blog’s tidbits from Altitude user conf:

  • cutting edge architecture (future-forward)
  • potential use cases: response stitching, dynamic video manifest, last-mile ad insertion
  • beta program is oversubscribed

Revenue growth dropped to 38% from 44% prior Q.

Muji, FSLY’s revenue is pretty seasonal with Q4 revenue typically the highest, so Q1 revenue 38% drop from Q4 44% is totally expected (only little Pandemic was factored in). We shall see the similar trend in 2020 as well assuming Pandemic traffic continues in H2 which implies Q4 2020 revenue may blow our mind.




muji - thanks for excellent recap.

I did have more % in NET before I understood FSLY and then I switched, thankfully mid-way through its tremendous runup…
(however, they are both low single digit % for me, just because I am not ready to cash out from LVGO, CRWD or DDOG to deploy into FSLY or NET)

One potential difference that can show up in upcoming earnings is that FSLY expectations are higher and NET expectations are moderate… AND recall NET offered 60 days (or 90 days? I forget) free for Cloudflare for Teams which was signing up many customers… so even if their revenue comes in line, probably NET Sept Q guidance may come in elevated compared to expectations…

Although both are priced are around 35 PS right now, expectations vs guidance can make a meaningful relative uptick in short term performance for NET… (unless ofcourse FSLY comes in with even bigger upside either on revenue or gross margins…)

I am still keeping both at the current relative ratio of my holding but tempted to buy some calls on NET to play earning…

1 Like

One potential difference that can show up in upcoming earnings is that FSLY expectations are higher and NET expectations are moderate… AND recall NET offered 60 days (or 90 days? I forget) free for Cloudflare for Teams which was signing up many customers… so even if their revenue comes in line, probably NET Sept Q guidance may come in elevated compared to expectations…

They mad it free for “at least” 6 months. If it ends in September as per the initial plan, we should see some benefits in Q3 and even more in Q4.…

I wanted to share a few thoughts and instead of creating another thread, I’m replying to this thread started by muji; so that relevant information about FSLY stays in one place. ( Also, thanks to muji for the reference to one of my earlier post on Fastly vs Cloudfare in his post.)

Well, before talking about FSLY, let me hijack this thread to share some insights about two of our favorite companies discussed on this board.

_A recent Zoom experience…_I attended a HashiCorp training to learn about their integration with Azure. It was a nice learning experience but it was a Zoom call with over 250 attendees that stood out to me. The call was divided into expert lectures and instructor lead lab sessions. For the lab sessions, the attendees were divided into groups of about 25 and were moved into “breakout rooms with a coach…which I felt was a pretty cool feature. I had attended an in-person HashiCorp training about a year back and the format was similar. The ONLY difference this time was that we were on a Zoom call. I think everything went smooth and easy :). The next logical question is… will companies continue to hold costly in-person trainings vs doing them online once things are back to normal ( if normal returns)?

…And if you ever pondered why Zoom choose Oracle Cloud instead of AWS…the answer is pretty clear. It’s NOT compute or storage. It’s the networking costs…That realtime video transfer cost is way higher in AWS ( not sure about Azure) compared to Oracle Cloud. Now isn’t that a brilliant move on the part of Zoom and should translate to some nice bottom line savings for Zoom…maybe 3-4x…who knows :slight_smile:

BTW, during my interactions with HashiCorp and FSLY, one company that was mentioned a number of times was DDOG. So, DDOG surely appears to be doing something very important and meaningful at the right time!

Anyways, coming to FSLY now…

If there’s a CDN and edge Cloud company that is really innovating it is FSLY!. No matter what I read and learn about the companies in this space, FSLY seems to be in a league of it’s own ( as far as research, building and innovation is concerned). Here are two great reasons why FSLY might be what CDNs and Edge cloud computing might look in the not very distant future. These may be crucial factors in how they stay ahead in their game and dominate edge cloud computing and CDN in the coming years.

1. CRDTs (Conflict-free replicated data type): In distributed computing, a conflict-free replicated data type (CRDT) is a data structure which can be replicated across multiple computers in a network, where the replicas can be updated independently and concurrently without coordination between the replicas, and where it is always mathematically possible to resolve inconsistencies which might result. If FSLY can express state at the edge in terms of CRDTs, it would enable local read and write transactions that stay very well within their latency bounds. According to FSLY, "CRDTs represent the most promising direction for writable state at the edge. and Fastly is actively prototyping and deploying a CRDT-based edge state system".

2. WebAssembly: Fastly build their native WebAssembly compiler and runtime to extend WebAssembly beyond the browser to create a platform for faster, safer execution on Fastly’s edge cloud and open sourced it ( Lucet). This should help developers focus on logic at the edge. I also see this feature forcing devs to rethink and decompose apps to move more logic that is better off at the edge.

On my personal investments…Like muji, I’ve been reducing my OKTA ( and a little bit of NET) position and increased my FSLY allocation in the past few months.

This is the difficult part: I think I’m getting a little biased and am seeing FSLY’s edge computing a step ahead of NET ( a lot of this bias may be driven by the developer side of me). Though I still have some of my NET shares, it’s a much smaller position compared to FSLY. It’s not surprising that some of the most innovative and developer focussed companies are using FSLY! Even Amazon is supposedly using Fastly’s CDN for their own sites, versus their internal CloudFront product.




For those of you who are technically inclined…

Google Cloud Platform Podcast with Tyler McMullen , CTO at Fastly,…

Supercharge your serverless with CRDTs at the Edge:…