Thursday night, CDN provider Fastly will price their shares and go public Friday morning under the ticker FSLY, looking to raise around $170M, with shares expected to price between $14-$16 a share. At the higher end of the price range, it would value Fastly at around $1.4B. The company has raised $219M to date and ended 2018 with 449 employees.

Fastly had $144.6M in revenue in calendar 2018, up 37.8% from its $105M in revenue in 2017. The company lost $30.9M in the 2018 calendar year, down 4.7% from $32.5M in losses in 2017. As of December 2018, 84% of Fastly’s revenue is derived from 227 enterprise-sized customers with each of those customers accounting for $530,000 in yearly revenue. [S-1 Filing]

On Wall Street, the biggest question I’m being asked is whether or not Fastly can become a $500M company in 3 years and hit $1.5B in revenue within 10 years, while also increasing margins in the 70% range. While it’s too early to be able to answer the question, based on what I see in the market I think the 3-year target for Fastly is doable. I closely track the size and growth rate of the markets Fastly is in, talk to and survey over 1,000 customers a year who use these services and I see the use cases that are coming with compute at the edge. 

All of this data adds up to a much bigger growth opportunity for these services over the next few years as more customers demand greater flexibility and better performance across their entire IT business stack. I’ve also known Fastly as a company for a long time, and am impressed by how disciplined they have been over the last few years from an operational level, across product, sales and marketing. The company has done an amazing job over the last 2-3 years disrupting on performance, ease of use, flexibility, and price. But make no mistake, Fastly isn’t selling on price. The company is winning on performance, the functionality of their services, being extremely quick and nimble, and simply doing it all at a lower price than competitors. 

Fastly is very good at passing on business that isn’t strategic and/or is priced too low and they don’t sign contracts simple to “fill the pipes”. Fastly isn’t trying to have a product portfolio for every sized customer, in every vertical and for every use case. They are very strategic on how they roll out services into new markets and target certain sized customers with use cases that solve specific problems. As a result of Fastly’s success to date, most are naturally wondering what impact Fastly is having on Akamai as Fastly raises more money, and what that competitive threat looks like going forward as Fastly continues to grow. 

While Fastly is pressuring Akamai in many segments of their business, winning deals on performance, price, and ease of use, there is only so much business Fastly can take from Akamai each year, simply based on their size. By my estimates, Fastly gets exposed to about 10% more of Akamai based contracts each year. So Fastly isn’t going to take a huge percentage of Akamai’s business overnight. But that doesn’t mean they can’t disrupt the entire market by forcing other CDNs to do a better job competing on price and performance, which they are doing.

Fastly has done a great job growing their business past the $100M revenue mark, which is where most CDNs in the past have stumbled. But from day-one, Fastly’s been a different CDN, targeting and specializing in the web performance market, not video streaming. From a network standpoint, Fastly has built a very different underlying network architecture, and as a result, is adding capacity without anything close to the CAPEX costs others CDNs are spending. Fastly has no routers or switches and its strength is the fact that they operate a software defined network with a footprint of 1,596 servers (as of March 31, 2019) across 60 POPs. If you want to geek out on how Fastly deploys their infrastructure, check out this video presentation by them from last year.

A big difference between Fastly and Akamai is the fact that Fastly is one network, for all of their services. A single network. Akamai has many networks, all built for a specific use case, for instance video, TLS, etc. which makes it hard to achieve efficiency at scale when you are buying and operating so many networks for each purpose. But one of the most important aspects of Fastly’s network deployment is the fact that they are a programmable edge. And while everyone uses the term “edge” without ever defining it, here’s how to think of the uniqueness of Fastly. 

If a customer builds their own DDoS, paywall, or image optimization service, they can run it from the edge on Fastly’s network. Nearly every other CDN requires the customer to have to use their DDoS, or image optimization service and doesn’t support the running of third-party applications on their infrastructure. Running applications and compute at the edge, that’s where this market is headed, which is why Fastly calls themselves an “edge cloud”. And CDNs that don’t have that functionality at scale, with flexibility and performance, are going to struggle to grow their business as more customers in the market require the functionality. This is exactly why it’s an area Akamai has said they are “increasing their investment” in that functionality.

Fastly is on a great path to continued growth but we need to watch how they scale their business over the next few years. As they begin to enter new verticals like banking and finance, a vertical that Akamai downright owns, how quickly can Fastly grow their revenue in new markets? How much can they grow their wallet share with current customers and increase their margins? Fastly rightly deserves credit for what they have already done and the expertise they have proven in the market. But now their work really begins to take their business to a whole new level and scale their network, revenue and customer base, with the target of being profitable. They will be a fun one to watch.

Note: I have never bought, sold or traded a single share of stock in any public CDN company ever. I am not receiving nor buying any shares in Fastly. My stock purchases have been restricted to Apple, Netflix, Facebook and NVIDIA.

I’ve recently completed my bi-annual survey of customers taking media delivery services from third-party CDNs and the data shows pricing on very large deals is now down to an all-time low of $0.001 per GB delivered. That’s one tenth of one penny and is the lowest pricing I have ever seen in the market. [If you are interested in purchasing all of the raw data including pricing, traffic growth, contract terms, CDN(s) used etc. – minus customer names, please contact me.]

That’s not to suggest that all customers for media delivery services involving video streaming and software downloads are paying one tenth of a penny per GB delivered, some are still in the $0.003-$0.005 price range per GB delivered. But for the largest customers, $0.001 per GB delivered is now the lowest price in contracts. With customers having so many different CDNs to choose from, both regionally and globally, and the ease of use of switching between multi-CDN vendors, media delivery simply isn’t a sticky product. Customers have a lot of choice and flexibility when it comes to video streaming and other media services and all of the mid-tier and larger customers have been using multiple CDNs for some time, with that trend growing.

Amongst all the major CDN providers, pricing for media delivery services is pretty on-par with one another, especially on new deals, and some CDNs, in particular Fastly, are passing on deals where the price is too low. Amazon Web Services is still very aggressive on price and loves to keep pushing pricing down, but customers need to be wary of Amazon’s egress charges which sometimes, can be larger than the CDN bill itself. [See my post here on that topic]

At the $0.001 per GB delivered price point it leaves almost no room for any CDNs to be profitable purely based on that traffic alone, so CDNs are finding other ways to squeeze out some profits. For example additional fees around low-latency, HTTPS, and live events. In most cases, for large one-off live events, customers also have to pay an RSVP fee for capacity, even if they don’t use it and they are charged based on a per Tbps sustained model. That’s not to say live events are a big revenue driver for CDNs, they aren’t, not even the Olympics or the Super Bowl.

Most large scale one-day live events are worth at most, a low six figures to any one CDN for the video delivery and don’t really move the needle from a revenue growth standpoint. But the point is that CDNs bundle media delivery in with other services and in some cases can charge certain customers for other pro-services help. The largest CDN customers don’t buy based on per GB delivered but rather on a per Mbps sustained model and many times, that is bundled in with other non-media services all wrapped up into one price per Mbps, across all of the vendor’s services.

Of course at the $0.001 per GB delivered price, many are going to suggest doom and gloom for the CDN vendors in the market, since pricing is still falling and Amazon Web Services continues to be so aggressive overall. While CDN vendors are still needed in the market and will continue to grow, it will be slow when it comes to OTT video delivery in particular. The largest growth of video on the Internet is being done by companies like Netflix, Facebook, Google, Microsoft and others who do all or most of the delivery themselves. Yes, we do have quarters where someone like Apple or Microsoft will give third-party CDNs more traffic, but it’s short lived. 

And even with new services coming out from Disney, Apple, AT&T and others, much of that will be delivered from the customer themselves. And the portion of traffic for new OTT services that does go to third-party CDNs will be split amongst multiple vendors. So if you think new OTT traffic is going to drive a big upswing in revenue to any CDN vendor quickly, you’ll be disappointed. Video traffic on CDNs is growing, but organically and it’s the product with the least amount of margin and the highest portion of CAPEX.

You also have some OTT services LOSING subscribers, as is the case with AT&T’s DIRECT NOW service, which lost 350,000 streaming subscribers in the last six months, which means less traffic to multiple CDNs, for that specific customer. AT&T is also currently in the stages of building out their own CDN for their WarnerMedia OTT product for those on AT&T’s network and I would expect that over time, some other new big OTT players might go the DIY route as well.

It’s also important to note that when CDN vendors say the gaming vertical is one showing good growth, it’s simply downloads of gaming content. No traditional CDN is doing the actual live streaming of multi-player gaming. That’s not something the CDNs offer and if you want to know all the details on how multi-player gaming video streaming is done, game companies like Riot and others have tons of technical details on their blog.  CDNs do help the gaming companies primarily with large software downloads and that’s still a great segment of the market for all CDN vendors, but third-party CDNs will not be “streaming” any video for Google’s Stadia cloud gaming service or the like.

Adobe Media Server (AMS), formerly Flash Media Server (FMS), and RTMP were always a bit ahead of their time.  Initially designed as a development platform for interactive, real-time communication (think Zoom or Webex circa 2003), the product ended up taking hold in low latency video streaming and helped birth the web video streaming industry. Today with interactive video and communication the core part of services such as Twitch, eSports and online betting, Adobe Media Server and RTMP may be headed for another day in the sun.

Under agreement with Adobe, a company called Veriskope plans to sell, support, upgrade, and enhance the Adobe Media Server (AMS), the RTMP SDK and several related video products worldwide. The company has entered into a global licensing agreement with Adobe to manage all aspects of Adobe Media Server business including global distribution, support, product development and operations. Veriskope is made of former Adobe and Macromedia alumni including Sarah Allen, Craig Barberich and Robert Pierce, whom worked on or led the development of Adobe Shockwave Multiuser Server, Adobe Flash Media Server, Flash video and Adobe Media Player.  

Thousands of education, government, enterprise and media companies continue to use the Adobe Media Server and the RTMP SDK for online learning, real-time communication, live sports content ingest, live event streaming and online corporate training. The partnership between Adobe and Veriskope will provide these customers assurance they can continue to use Adobe Media Server and RTMP into the future. Veriskope will also provide services to help existing customers modernize and migrate from legacy Flash experiences to new web native HTML experiences.  

Veriskope ‘s service offering for AMS customers includes software licenses, upgrades, support plans, and product enhancements for Adobe Media Server, RTMP SDK, Flash Media Live Encoder and the Adobe Media Gateway.  In addition, the company is offering professional services support including migration services to help customers quickly move from Flash UI experiences to native web HTML, engineering services to design and architect new solutions, and training to help companies more effectively utilize the products.

Veriskope says their immediate plan is to focus on supporting existing AMS and RTMP SDK customers with the short term goal of providing a low risk, cost effective solution that allows thousands of AMS customers to transition from a Flash user experience to web native HTML experience.  

So does this mean that Adobe Flash will continue? The exciting part about AMS and the RTMP SDK is that it does not require Flash. RTMP was designed to be open and support different types of client software. Today AMS supports HLS and in the near future, Veriskope hopes to make announcements about supporting other web native formats.  The company will continue to support customers who use Flash and AIR for on-device experiences in cars, on boats, on industrial devices and more. AMS is not Flash. RTMP is not Flash. AMS and the RTMP SDK provide scalable, interactive video infrastructure that works across the web, the cloud, the edge and more.

As you look a few years into the future there are many standard features of the AMS and the RTMP SDK products that are more relevant today than when they were introduced — time synchronizing video, messaging and data for gaming, eSports and betting; proven low-latency ingest and contribution for live events, elearning and user-generated content; video and AI merged on the smart edge in IOT, manufacturing and retail and last but not least, the future of interactive communications.

Going into YouTube’s presentation at Radio City Music Hall on Thursday night as part of the NewFronts, one would think that the main focus would be Google’s desire to convince advertisers to move their ad dollars to YouTube and YouTube TV. But you’d be wrong. The event was a star studded music event and barely touched on their TV business at all.

The only data YouTube gave out was commissioned by the company via Nielsen, or their own Google/MediaScience Lab, and they didn’t disclose how many subscribers they have to YouTube TV, which content users consume most, or what any of the trends are when it comes to live TV or on-demand content. They gave out a metric on reach, saying the YouTube platform reaches more 18-49 year olds in an average week than all cable TV networks combined, but actual data on engagement would have been much more valuable.

Only one advertiser, Johnson & Johnson, co-presented with YouTube talking a few numbers, but the rest of the time was mostly taken up by live performances by eight singers/bands. YouTube said watch time of YouTube on television screens tops 250 million hours per day, but didn’t break the number down past that or say how many of those hours consisted of ad based content.

The tag line for the event was “Prime Time Is Personal”, with no focus on TV at all, and nothing about the performers tied into the “personal” theme in any way. Overall, it was a very odd event with a big focus on music, original shows, new music specials and new learning series. The real only piece of news from the event was YouTube announcing that all of their original shows will stream for free, with ads.

Overall, the presentation and experience felt more like an event for fans, as opposed to a night of information targeting brands and advertisers.

While OTT viewership is on the rise, the number of streaming providers is also skyrocketing with Disney, Apple, WarnerMedia and many others launching platforms later this year. With competition as fierce as ever, and costs for content creation and licensing so high, OTT providers can’t afford to leave any monetization stone unturned.

To thrive in such a competitive environment, streaming companies need to ensure that they are providing a user experience that not only meets, but exceeds viewers’ expectations. With so many streaming apps in the market, users will have no problem ditching one service that plagues them with annoyances like buffering for one that offers a seamless experience. It’s impossible that every service will survive the long haul, and users will decide which video providers continue to reap the benefits of the growing OTT market.

One feature that promises to solve the most pressing issues faced by users, while also giving streaming providers new revenue opportunities, is mobile video downloads. Many of the common problems encountered by viewers are the result of connectivity issues, especially on mobile. Globally, many viewers are plagued by problematic cellular connections that can’t support their on-demand video needs. Studies show that viewers are increasingly impatient, and the lag or buffering that results from insufficient connectivity can create an intolerable user experience.

Some companies believe that 5G cellular technology will be a solution to these types of viewership issues. Just like 4G before it, 5G promises to be the holy grail of the mobile experience that solves users connectivity problems. However, 5G won’t be a solution in all situations. Not only will its eventual rollout be slow, especially in many places around the world, but it won’t be able to solve for simultaneous advances being made in video quality. Even when 5G ultimately comes about, there will still be an increased load on networks, especially as higher bitrates come about. Download, though, actually solves many QoE viewership issues by allowing users to download videos to their device when they have a great connection, so they can watch when connectivity is poor or nonexistent.

Yet download is more than just a way to make video viewing less frustrating for end users. It also presents an enormous revenue opportunity for providers. Of course, for SVOD providers, a feature like download increases engagement and reduces churn, which is a huge business win. But for AVOD providers, there’s an even more direct benefit of download: the monetization of downloads. Attaching dynamic ads to downloaded video gives ad-supported video providers a new way to reach viewers who otherwise couldn’t engage with content—and that means an increase of ad dollars.

When offline, AVOD viewers lose the ability to watch ad-supported content, which currently requires a wifi or cellular connection to view. That means millions of viewers can’t engage with AVOD companies at times when they would otherwise be watching video in high volume: on planes, commuting, on vacation, and other areas that have poor Internet service. With download, though, viewers can anticipate these moments and have content ready to view seamlessly when they’re on the go. And that means video providers can catch extra ad dollars that were previously unavailable.

Hypothetically speaking, if an AVOD service has 10M subscribers and 25% of them download one 30 minute video per month, with an average ad load of five ads per show, at a CPM rate of $20, that’s an additional $250,000 in ad revenue per month. And that’s with only 25% of the users utilizing the download option, and I think it would be much higher than that. In fact If you increased the usage of download activity to 1 hour per week, the revenue opportunity would be $2 million per month or $25 mil per year.

The business benefits of download can’t be overstated. Every downloaded video is new engagement earned, loyalty gained, and dollars won. Hulu has teased a potential rollout of ad-supported download feature in the past, but so far the AVOD industry has failed to prioritize this lucrative solution to lost viewership. Without ad-supported download, AVOD providers are leaving money on the table every day. And this holds true across the entire video streaming industry. Without features like download, providers risk losing views and subscribers to the services who allow viewers to watch whenever and wherever they want.