Twilio and Tin Plates
Twilio and Tin Plates
How a business lesson from the Carnegie Steel Company can help us understand the Cloud Wars.
Twilio and Tin Plates
Twilio ($TWLO) agreed to acquire Sendgrid ($SEND) earlier this week. The deal makes sense and there’s a point of view developing that this type of consolidation will continue:
But rather than celebrate, I encourage Twilio and similar web services companies to prepare for direct competition with the web service divisions of Amazon, Microsoft, and Google as the Cloud Wars continue to rage on.
The Cloud Wars will move up the stack
In a Techcrunch op-ed last year, Box founder and CEO Aaron Levie declared “now the cloud wars (really) begin,” predicting that as Amazon, Microsoft, and Google’s web service divisions duke it out on price, they’ll need to seek profits and differentiation through other means. I believe this differentiation will come from building specialized web services on top of the hosting, storage, and cloud computing infrastructure they’ve already built, essentially move up the stack to compete with companies like Twilio.
I don’t I mean to trivialize the quality of Twilio’s product. It’s easy to use, developers love it, and they’ve created a competitive moat by stitching together telephony services from over 180 countries. It’s no small feat to replicate all of this, but we must admit AWS has the appetite and growing acumen to do some damage here. AWS’s first move up the stack was in offering functionality like analytics, a feature several venture-backed startups are building. Amazon then built a mobile identity management suite into AWS with functionality similar to what Parse sold to Facebook for $80M a few years ago. Loftier aspirations are revealed in an Amazon job post from last year, which indicates AWS might be moving into payments soon. Yes, Amazon does have some expertise fighting fraud within their retail offering, but offering a white-label payments web service is at least as difficult as managing a multi-national web of telco integrations. And Amazon’s latest web service endeavor is the most sophisticated and ambitious yet. Alexa takes on the task of natural language processing with a voice recognition service that not only gets cheaper with scale (like AWS), but actually increases in quality the more developers use it. Companies like api.ai will find it difficult to compete here. I don’t mean to suggest that any web service is vulnerable just because it’s built on top of a hosting provider like AWS. Only certain developer-focused services are at risk and which ones depend on what moves and countermoves play out in the ongoing Cloud Wars.
So is Twilio a $7B+ market inefficiency or a business that provides enough value and defensive features to ward off competition from the likes of AWS on its own?
An inefficient quirk?
Zach Seward’s Quartz article from Twilio’s IPO last year calls Twilio’s “the most interesting tech IPO of the year,” then goes on to cite the company’s utter dependence on AWS:
[Twilio seems] less like SaaS and more like a cloud hosting service. The best comparison is perhaps to Amazon Web Services (AWS). Nowadays, most startups use AWS or its competitors for hosting instead of worrying about administering servers themselves. It’s cheaper, more reliable, and easier to scale than going it alone. So if Twilio is akin to AWS, then here’s the really fascinating part: Twilio hosts “substantially all” of its own service on AWS.
Twilio customers, in other words, are outsourcing messaging to Twilio, which in turn outsources to Amazon. That’s either an inefficient quirk, or it points to a new kind of company in the startup stack, a layer in between infrastructure and software. Call it AWS as a service — AWSaaS?
Mr. Seward interprets the business model of companies (e.g. Twilio, Stripe) building specialized web infrastructure on top of AWS as a huge opportunity–at worst it’s an “inefficient quirk.” I see Twilio’s potential competition against AWS as a significant risk. For Twilio, competition with Amazon isn’t just intimidating because Jeff Bezos loves to build low-margin, capital-intensive businesses and has no problem giving away profits to obtain market share (AWS has reduced its prices 49 times since it was launched in 2006). The math of competition just doesn’t work in Twilio’s favor:
Twilio can’t outperform their primary web service provider on server uptime and availability–a critical aspect of any web service–and you can infer from the fact that they are so dependent on AWS that Twilio can’t go elsewhere for similar hosting functionality at lower cost.
Twilio can’t charge less than AWS for their specialized messaging services because AWS can likely negotiate similar (or cheaper) telephony rates based on their reputation.
AWS has 1M active customers to Twilio’s 28K. That scale advantage allows AWS upsell existing customers on specialized messaging services negotiate telephony rates down even further.
Compounding this threat is the fact that Twilio really only needs to lose one large customer for this risk to become existential. Twilio is highly dependent on WhatsApp in particular for revenue and three of their top ten customers aren’t locked into long-term contracts. From Twilio’s S-1:
In 2013, 2014 and 2015 and the three months ended March 31, 2016, WhatsApp accounted for 11%, 13%, 17% and 15% of our revenue, respectively…Our Variable Customer Accounts, including WhatsApp, do not have long-term contracts with us and may reduce or fully terminate their usage of our products at any time without penalty or termination charges.
AWS really just needs to build a good enough text messaging or VoIP offering as part of AWS and outsell Twilio based on price and access to developers through their existing web hosting relationships to be dangerous.
A lesson from Carnegie
Twilio should just move off AWS to another provider or to build their own hosting infrastructure to mitigate the risk of competition, right? Well, US business history provides a lesson.
In early 1899, JP Morgan rolled up over 85% of the finished steel manufacturers in the country (producers of metal pipes, wire, and tin plates) into the National Tube Company, then struck deals to source National Tube’s primary steel exclusively from competitors of Carnegie Steel, the largest primary steel manufacturer in the country at the time. This maneuver stole market share from Andrew Carnegie’s core business and he responded by forming a finished steel manufacturer of his own, which could always sell its finished products for less than National Tube. From Charles R. Morris’ The Tycoons:
Given the relatively modest investment required to enter most finished steel businesses, it would always be easier for the primary steel companies to integrate forward into wire, hoops, or tubes. And by using their new finished lines to sop up surplus primary steel capacity, they would have the luxury of selling below cost and killing off independents at will.
In our modern-day web services analogy, AWS is Carnegie Steel, Microsoft Azure and Google Cloud Platform are its competitors, and Twilio is a tin plate manufacturer. Twilio can’t integrate backward into a wholesale web service provider. And if enough customers like Spotify and Apple continue to leave AWS for Google Cloud Platform or Azure, we might see AWS start to go on the offensive.
Beat ’em or join ‘em
There’s a silver lining in the Cloud Wars for some web service companies looking for liquidity. As Google Cloud Platform and Azure look for ways to compete with AWS they may become acquisition-happy.