Welcome to Mostly Cloudy! Today: why financial engineering is becoming more important than systems engineering, how the chatbot frenzy is fulfilling the wildest dreams of the spam merchants, and are we ready for data centers on the moon?
Photo: Gaudenis G. on Unsplash
The CFO strikes back
Over the past decade a lot of technology purchasing decisions shifted away from a top-down, finance-department driven procurement process to one controlled directly by business units, who are generally in the best position to know what they actually need and what actually works. That was a good thing: it unleashed a ton of innnovation as cloud computing startups honed in on specific needs and allowed buyers to experiment with pay-as-you-go models until they found the right match.
But economic worries are forcing everyone to scruntize their spending, which means the finance department has regained a modicum of control over the buying process. It’s hard to see this world regressing back to the days of massive one-size-fits-all enterprise software packages sold on the 11th fairway, but in the meantime, the FinOps opportunity has never been stronger.
Vantage is the latest startup to tackle a fundamental problem of operating enterprise technology services at scale in the 2020s: understanding exactly where, and how much, your company is spending on those services. It announced a $21 million round of funding Wednesday to help customers manage spending not only on cloud infrastructure providers but also application companies like Snowflake, Datadog, and Twilio, among others.
The decentralization of technology spending helped buyers move faster, but it also led to companies losing track of cloud instances that were spun up for a skunkworks project and never turned off. Likewise, changes in demand caused by things like global pandemics forced those companies to quickly embrace new SaaS tools while still paying per-user prices for older tools they weren’t actually using.
FinOps vendors promise to help buyers track down all their spending (harder than it sounds) and figure out what they need to prioritize. It’s very easy to over-provision cloud infrastructure services for apps that don’t really need that much horsepower, especially if cloud customers skipped the step of reconfiguring apps built for on-premises data centers to run efficiently on cloud services.
Cloud customers have been complaining about how easy it is to overspend on the cloud since the earliest days of the cloud, but there are more tools available than ever before to help them control that spending. Still, there’s no question at this point that the cloud providers themselves could — and should — be doing a much better job of helping their customers understand their spending habits, especially those who have migrated to the cloud in just the last few years.
While some disgruntled cloud customers are throwing in the towel and rolling their own data centers again, those worried about their cloud spending habits might be better off starting with a FinOps assessment before going down that path.
We must not allow a chatbot gap
With another week arrived another cascade of companies scrambling to add OpenAI’s ChatGPT to their enterprise products, whether or not they understand how it works or where it makes sense. However, one clear use case is emerging: spam.
According to CNBC, Salesforce and other marketing-centric companies are using ChatGPT to help customers spin up cold email pitches to potential customers. One video of a ChatGPT-enabled spreadsheet auto-generating pitch emails apparently went viral on TikTok, which is a little depressing.
To be sure, it’s not like the past 25 years of spam email has been distinguished by its personal touch. But it’s a little surprising that corporate marketing teams would so readily turn their customer interactions over to a chatbot that is not exactly known for accuracy, which means somebody will have to sit there and make sure the chatbot output adheres to the branding guidelines anyway.
The real question: Will Microsoft and Google allow Outlook and Gmail to detect and sort spam email written by customers using their respective enterprise chatbot products?
Around the enterprise
Lonestar Data Holdings received $5 million in funding to put data centers on the moon this year, bringing new meaning to the term “cold storage.”
Datadog suffered a massive, hours-long outage Wednesday that was still ongoing as of this writing and particularly damaging for a company that makes money monitoring systems for customers trying to prevent outages.
Twitter also didn’t work for several hours on Monday, a regular-enough occurence at this point to start letting ChatGPT write the summaries.
If and when quantum computers arrive in a useful state, Microsoft outlined its plans to use powerful classical computers to work in tandem with quantum machines to stablize their output.
Thanks for reading — see you next week!