Hello, and welcome to Protocol Enterprise! Today:
why reports that Elon Musk wants to dramatically cut Twitter’s infrastructure spending could damage the company as much as anything else he’s done, AWS CEO Adam Selipsky on hiring priorities, and the tao of computing.
Execution errors
Let’s get one thing out of the way up front: Twitter has always been one of the most obvious examples of one of the lesser-known aspects of the tech industry, that the hardware and software that powers some of the world’s most important and influential online services is often held together through a series of daily miracles and sheer gumption. But the wrecking ball that Elon Musk sent through Twitter this week could easily upset that delicate balance and end the company faster than any advertiser boycott might.
According to Reuters , this week Musk directed Twitter engineers to slash $1 billion from the company’s annual tech infrastructure budget by Monday, before laying off thousands of employees Friday. Given that Twitter reported $1.8 billion as its cost of generating revenue for its fiscal 2021 year — infrastructure costs are a hefty portion of that figure, but not the sole contributor — if that number is accurate, we’re talking about enormous cuts.
We know a little bit about Twitter’s current infrastructure strategy .
- Like many companies born in the mid-2000s before cloud computing had really matured, Twitter initially ran on self-managed data centers.
- Unlike many companies born around that time, Twitter was notoriously unreliable in those early days, routinely going down during sporting events and Apple keynotes and giving birth to the infamous “fail whale.”
- However, Twitter engineers were able to come up with unique ways to solve those reliability problems, leading to the birth of now widely used concepts like the service mesh .
- While the company still operates its own data centers, in 2018 it moved a big chunk of its data infrastructure to Google Cloud, and in 2020 it signed a multiyear deal with AWS to run the actual real-time tweet timelines on the cloud leader’s servers.
One does not simply break a multiyear computing infrastructure deal with AWS, especially over a weekend.
- While talk of “cloud repatriation” is hot among venture capitalists tired of spending startup money on cloud services, the canonical example of modern cloud repatriation — Dropbox — required several years of work before the company felt comfortable with the switch.
- It’s reasonable to assume that Twitter, like many companies relatively new to the cloud, has over-provisioned the capacity it actually needs and can find some cost savings with a hard look at its spending.
- But if half the people who understand Twitter’s workload requirements just got kicked out the door, those who remain will be forced to make hasty decisions about what can stay and what can go without fully understanding the ramifications of those decisions.
Musk’s operational challenges are clear: He needs to cut costs to service the $1 billion in annual debt payments he saddled the company with by taking it private.
- But as we learned from Mudge’s whistleblower report , Twitter’s infrastructure was already creaky and lacked some of the backup and recovery options that are considered table stakes for businesses operating internet services of its type.
- That means any disruption to the Rube Goldberg machine that lets the tweets flow could simply make Twitter unusable for long periods of time.
If the report is accurate, slashing Twitter’s infrastructure costs nearly in half overnight will immediately affect the stability and reliability of the service.
- I mean, it’s not rocket science.
Sponsored content from ServiceNow
Ever since the pandemic put the evolution of everything in hyperdrive, marketers realized the old categories of B2B, B2C, and B2B2C were obsolete. Starting in 2020, our profession embraced the Business to People (B2P) paradigm. Business, fundamentally, is relationships among people. Even in the biggest enterprises, those making momentous decisions are still people.
Selipsky’s ‘very conservative’ AWS
AWS is ratcheting down its hiring for new positions at the cloud computing provider, according to CEO Adam Selipsky.
“AWS has done a lot of hiring to drive innovation and work with customers over the past few years,” Selipsky told Protocol in an interview on Friday. “We’ve grown significantly. We have, I think, a strong set of resources. We will definitely slow down our growth … in hiring.”
The news follows word this week from Beth Galetti, senior vice president of people experience and technology at parent company Amazon, that the retail and tech giant will temporarily halt new incremental hires for its corporate workforce due to “an unusual macroeconomic environment,” but will continue to hire in “targeted places.”
Galetti said Amazon wants to balance its hiring and investments with being “thoughtful” about the economy.
“With the economy in an uncertain place and in light of how many people we have hired in the last few years, [Amazon CEO Andy Jassy] and S-team decided this week to pause on new incremental hires in our corporate workforce,” Galetti said in a message shared with employees on Wednesday and made public on Thursday. “We had already done so in a few of our businesses in recent weeks and have added our other businesses to this approach.”
“Across Amazon, we’re going to be very conservative just in the immediate future about the resources we bring on board,” Selipsky said. “AWS will also be very conservative about new resources we bring on board. We’re always concerned with the long-term health of the business. And if there’s something that we need to do to serve customers or build an important capability, we’ll take the long-term view.”
We’ll have more from Protocol’s in-depth interview with Selipsky in the coming weeks. Stay tuned.
An invite to ponder planetary computation
What is the future of computation? How will tech stacks affect the geopolitical order in years to come? Is the earth gradually developing its own intelligent awareness?
If these are the sorts of questions that excite and inspire you, a computing philosophy project launched by the Berggruen Institute — which will pay to convene philosophers, designers, technologists, and other techno-thinkers in Los Angeles, Mexico City, and Seoul to ponder them — is seeking program participants.
“The goal really is to shift a theoretical and practical, philosophical discourse around computation that will reorient computation toward a more productive relationship to planetary futures,” Benjamin Bratton, professor at the University of California, San Diego, and director of the program, told me last month. “Computation is a necessary part of that equation,” he said.
Take climate change. Bratton said, “The very idea of climate change is itself an outcome of planetary-scale computation. Without the sensors and simulations and super-computing models, the very idea of climate change, at least at its scientific granularity, can’t exist.”
The Antikythera program, named after the Antikythera mechanism — the world’s first known computer – is accepting applications through Nov. 11.
— Kate Kaye ( email | twitter )Around the enterprise
Microsoft said the percentage of cyberattacks by nation-state groups targeting critical infrastructure reached 40% during the 12 months ended in June 2022, doubling year-over-year, driven by Russia-linked attacks on Ukraine and espionage against the U.S. and other Ukrainian allies.
Alibaba Cloud will use its internally developed Arm server chips to power 20% of its instances by 2025, the company said this week.
Sponsored content from ServiceNow
The pandemic has been a global event that, somewhat paradoxically, put an intense spotlight on the personal. In a marketing context, it underlined the centrality of supporting customers’ purpose – personal and organizational – and the need to serve the customers’ customer hierarchy of needs as those needs change over time.
Thanks for reading — see you Monday!