Welcome to Mostly Cloudy! Today: Salesforce beats earnings expectations and puts its own salespeople on notice, the evolving definition of serverless at AWS, and OpenAI reveals more of its enterprise plan.
Photo: Salesforce
From such great heights
With hindsight — a term that gets a bad rap but really means “here’s what actually happened” — the 2018 opening of the Salesforce Tower in downtown San Francisco was the peak of Salesforce’s run as an enterprise software wunderkind. After spending almost $50 billion since then on the acqusitions of Mulesoft, Tableau, and Slack in a bid to keep Salesforce growing while core products such as Sales Cloud and Service Cloud matured, Marc Benioff has scrambled to keep control of the SaaS pioneer in 2023.
On Wednesday, he bought himself significant breathing room.
Salesforce’s fourth-quarter earnings beat expectations on all fronts, and the company raised its full-year outlook. Investors pushed the company’s stock up as much as 15% in after-hours trading, a welcome respite for Benioff after a rough year thanks to the broader pullback in enterprise SaaS stocks and the company’s self-inflicted woes.
While the investor push inside Salesforce has focused on its costs and sales efficiency, the company still faces an important challenge; it doesn’t have a differentiated product strategy for a maturing enterprise software market. Salesforce is a service that companies keep renewing because they kind of have to, after years of recording customer and prospect data in cloud services led by the company’s collection of furries.
But as Aisha Counts and Joe Williams pointed out last year at Protocol Enterprise, enterprise customers increasingly want to use Salesforce data outside of Salesforce’s core products, as part of a broader set of business data from several sources that can be used to make decisions. After Apple allowed iPhone users to opt out of third-party tracking services, enterprise vendors and marketing departments rushed to build first-party customer-data platforms of their own in order to make sure they understand what their customers need and want.
In other words, Salesforce data is still valuable to its customers, but they’re finding software that surfaces that data alongside other sources of corporate and customer data more valuable. That led to the development of Genie, or the Salesforce Data Cloud, a cross-Salesforce data layer that Benioff spent several minutes extolling to the usual fawning group of financial analysts allowed on a conference call following the release of the results.
However, chaotic integration processes following years of Salesforce acqusitions have made it much harder to surface that customer data across the breadth of Salesforce’s holdings, and customers become frustrated rather quickly when they can’t manage their data their way. Benioff rattled off several early customers for its new data platform during the call, but it’s far from clear that Salesforce is a leader in this emerging market, one of the hottest areas of investment across enterprise software.
Wednesday’s results will please Salesforce’s activist investors, who have already made a tidy profit. But during the conference call Benioff signaled pretty clearly where Salesforce is headed over the next several years.
He revealed that the first person who texted him congratulations after the earnings release Wednesday was Oracle co-founder Larry Ellison, his longtime mentor, who has “given me the Oracle playbook” over the last few years. Given Oracle’s checkered reputation with its customers over the last decade, it sounds like Salesforce and its customers are entering a new, more confrontational period of their relationship as profits become paramount and Salesforce’s sales force comes under intense pressure from Benioff to sell more stuff.
It’s all Greek
One of the best running AWS jokes over the past few years has been the company’s evolving definition of “serverless computing,” which when it introduced its Lambda service way back in 2014 very clearly meant “functions as a service” and has since come to mean several different things depending on the context. Corey Quinn noted Wednesday that even Lambda’s definition of serverless has become warped over the years.
Quinn pointed out that Lambda currently supports only version 3.9 of the managed Python runtime and hasn’t updated that runtime to support several new versions of the Python runtime that have been released since August 2021. Python is one of the most popular programing languages on Stack Overflow’s annual survey of developers, so it’s a little surprising that AWS has dragged its heels when it comes to Lambda, which it has declared the future of its computing strategy on multiple occasions.
Lambda customers that want to use the latest Python runtime can configure it themselves, but as Quinn wrote, the whole point of Lambda is “that it removes undifferentiated heavy lifting for customers.” The clock is ticking: the 3.10 version of the Python runtime will enter its “security only” updates phase later this year, which leaves Lambda customers running Python in a rough spot.
Around the enterprise
OpenAI launched its enterprise API and new capacity plans Wednesday, while making a separate promise that it won’t use customer data to train its models without opt-in consent.
Snowflake reported a 54% jump in revenue that exceeded Wall Street expectations, but the investor class didn’t like its reduced outlook for the upcoming year and punished the stock in after-hours trading.
Dish Networks finally acknowledged that a ransomware attack took down its systems for several days, after initially keeping quiet about the causes of the massive outage.
Zoom beat Wall Street’s sharply reduced estimates for revenue growth and also topped profit expectations, promising better growth ahead as customers carefully scruntinize enterprise software spending.
Twitter had another outage for some reason.
Thanks for reading — see you next week!