ForwardThinking 🤔

Oracle's Tuesday morning massacre

On Tuesday morning, thousands of Oracle employees were sent packing with an email. While there is no great way to conduct a mass layoff, this move felt particularly cold, waking up to an impersonal missive with the line: "We are sharing some difficult news regarding your position." 

It was surely an awful morning for the estimated 20,000 to 30,000 employees who got that email — a number Oracle has yet to confirm or even acknowledge publicly. The company has been chasing AI infrastructure deals, particularly with OpenAI. It's an undertaking that has cost billions and generated mixed results to this point with its stock down over 25% so far this year. Even big layoffs, which counterintuitively usually appeal to investors, didn't move the needle much this week.

When you're dealing with surging capex, the money has to come from somewhere, and it seems to have fallen on the rank and file employees. Surely, Larry Ellison wasn't going to fire himself or perhaps take out a home equity loan on his private island, was he? 

Just how much has capex increased? Consider that in FY2024 the company spent roughly $7 billion. In FY2025 that jumped to more than $21 billion, and it’s projected to reach around $50 billion this fiscal year. That’s an astonishing spending trajectory.

It's also why this week's action didn’t come as a surprise. In fact, reports earlier this year predicted layoffs of this scale, with TD Cowen estimating Oracle could save up to $10 billion in employee costs to reinvest in capex. Talk about robbing Peter to pay Paul.

Customers should worry

Whatever the actual number ends up being, when you lay off this many people, especially at a company that lives in the enterprise, it raises serious questions about whether Oracle is sacrificing customer experience to bankroll a massive infrastructure push whose returns may not materialize for years, if they ever do.

While the remaining employees deal with survivor's guilt and ballooning workloads, enterprise customers have to be wondering how this will impact their relationship with Oracle, and how well the products, already perceived as expensive and difficult to operate, will be impacted by cuts this deep.

Matt Stava, CEO of Spinnaker Support, a company that provides third-party support services for Oracle, was wondering the same thing. "The risk is that core support services, which enterprise customers rely on to keep legacy systems stable and secure, become strained at the very moment complexity is increasing,” he said.

It's not just Oracle

Big layoffs are everywhere these days with Oracle being just the latest. It’s not hard to draw a line between rising costs and workforce cuts, and increasingly that line leads back to AI-related expenditures. 

The closest analog is Amazon, which laid off 14,000 people in October, then cut another 16,000 in early 2026 after projecting $200 billion in capex this year. Last year Microsoft laid off 15,000 employees including 6000 in May and another 9000 in July after spending over $88 billion in capex.

Over the last year, Meta has laid off over 5000 employees and a recent Reuters report suggests another monster layoff could be coming with up to 20% of the workforce or around 16,000 employees, potentially at risk. The company has been investing heavily in both infrastructure and high-end AI talent.

Google hasn't had the same numbers since laying off 12,000 in 2023, which was probably due to pandemic overhiring, but has continued to conduct smaller more surgical layoffs involving thousands of people over the last couple of years, while investing heavily in infrastructure.

Tech vendors will surely attribute this to AI replacing workers, but the more plausible story is that companies are cutting people to pay for their massive AI infrastructure binge. It’s a lousy trade‑off for the tens of thousands laid off to fund a bet that might never pay off.

The problem for all those workers is that whether it works or not, the bill just came due and they were sacrificed to pay it.
~Ron

What's new on the blog 📰

CrowdStrike CTO faces two-way challenge when it comes to AI

CrowdStrike CTO Elia Zaitsev was initially excited by the possibilities of large language models until he realized the attackers would have the same tools. I talked to him about the two-way challenge of building AI defenses while the adversaries use AI to rewrite the rules.

"Once the initial euphoria wore off, I started thinking, man, the adversaries are going to be using this stuff too."

Get the full story>>

Vibe coding won't kill SaaS, but complacency could

Everyone's talking about vibe coding killing SaaS, but the enterprise leaders I spoke to from Zuora, ServiceNow, LaunchDarkly, Forrester and SAP say it's not that simple. The real risk isn't where most people think it is.

"There have been a lot of lessons learned, and those lessons can't be vibe-coded."
~Cameron Etezadi, CTO, LaunchDarkly

Get the full story>>

Databricks takes on security and a whole new buyer

Databricks leapt into the cybersecurity market this week. While it technically makes sense to think about security as a data problem, the question is whether the company can shift buyer personas successfully.

Former Gartner analyst Sanjeev Mohan says it will require a different sales motion: "Selling to infosec, DevOps and SRE teams is very different from selling to analytics or business users. That's something they would need to develop."

Get the full story>>

How one person is securing all of Amazon from front door to firewall

Amazon is a huge company and when you consider subsidiaries like Whole Foods and MGM Studios, it's even bigger. That's why it's so hard to believe that one person is in charge of integrated security across the company for everything from badges to cyber.

His view on how most companies handle it? "I'm really not interested in taking a colander, blocking a few of the holes and calling it a bucket. That's called security theater."

Get the full story>>

As Narayen steps down, Adobe must confront its next big transition

Earlier this month, Shantanu Narayen announced he was stepping down as Adobe CEO after 18 years on the job. He oversaw some major transformations in his time in the corner office, including the move from boxed software to cloud subscriptions, but with his departure, the company may be facing its biggest transition yet.

"He did the impossible by saving and transforming an on-premise company into a real SaaS business. The book is yet to be written on the second transformation — to AI."
~Anshu Sharma, co-founder and CEO at security startup Skyflow

Get the full story>>

News of the Week 📣

Box transforms content management again, this time with agents

Box blew onto the enterprise content management scene more than two decades ago, looking to shift the industry from on-prem dinosaurs to modern cloud-based systems. It was a tall order, but today just about all content management is in the cloud. Now it's pushing further with the Box Agent, among the first general-purpose AI agents from major ECM vendors for executing workflows in plain language.

The company released the Box Agent on Thursday. Box CEO Aaron Levie says the agent has access to your entire Box environment, so you can describe a task, and it can access any files you have access to in the service of executing the work.

"So the power is now all of a sudden, it can take all of that enterprise content that we've amassed over the years, and instead of a person being a dependency to find a piece of information, the agent is able to now go and work and operate on your behalf, to go pull that information in and then it lets you execute your tasks," Levie told FastForward.

Levie says the benefit of the Box Agent is how well it understands the Box environment. "So it knows how your files are organized, it knows the people you're working with, it knows how to work with information, because it's just hyper wired to enterprise content," he said.

The Box Agent is now generally available.

Freshworks brings FireHydrant incident management into the fold with new integration

Freshworks announced this week that it has natively integrated FireHydrant, the incident management startup it acquired late last year, into its Freshservice platform. The goal is to unify IT service management and incident response into a single workflow.

"The idea is to sell our Freshservice offering in combination with FireHydrant starting this week," Srini Raghavan, chief product officer at Freshworks told FastForward. The company believes that combining the two offerings will deliver a more connected and comprehensive approach to managing all aspects of IT.

"When incidents occur there might be an engineer on‑call and a team facilitating the response through FireHydrant, but the system of record is your Freshservice platform," said Robert Ross, FireHydrant founder, who has taken on the role of senior director of product management since joining Freshworks.

Freshworks shifted away from its SMB roots in recent years, mostly shedding the marketing and sales part of the business in favor of a new focus on customer and employee experience. The customer focus is now between 2000 and 20,000 employees, what it calls mid-market.

FireHydrant was founded in 2018 and raised over $32 million across three rounds, per Crunchbase. Investors included Work-Bench, Menlo Ventures and Harmony Partners.

GitHub will use your Copilot data to train AI models unless you opt out

In what can only be described as a surprising move, GitHub announced in a blog post this week that it would train on user data unless they explicitly opted out to do so.

"From April 24 onward, interaction data—specifically inputs, outputs, code snippets, and associated context—from Copilot Free, Pro, and Pro+ users will be used to train and improve our AI models unless they opt out," GitHub's Mario Rodriguez wrote in the blog post.

While there has been a general move towards sharing prompt information in consumer AI, it's still unusual to train on paid accounts, especially developer code. One of the reasons that developers are reluctant to share data, code or other content with models is exactly this. They want to keep their code to themselves for obvious reasons.

It's worth noting that Copilot Business and Copilot Enterprise users will be exempt from this edict, but it only adds to the confusion around ownership that is likely to occur. Most people will just hear that GitHub is using your data unless you tell them not to and won't be paying attention to the differences between various licenses.

In general, it strikes me as a terrible idea, and one that could adversely impact the brand perception with regular customers.

What I'm reading 📚

Teens Sick of Their iPhones Are ‘Mallmaxxing’
~By Lily Meier, Bloomberg

The Decadelong Feud Shaping the Future of AI
~By Keach Hagey, Wall Street Journal

What the heck is going on with Microsoft lately?
~By Mary Jo Foley, GeekWire

Wikipedia bans AI-generated articles
~By Emma Ross, The Verge

Look who's talking 👄

“I fundamentally believe that AI is going to augment and accelerate how we work, and it's not going to wipe out all the jobs. And I don't think it's going to become Terminator. So we're very much on the optimistic side of what work is going to look like.”

~Box CEO Aaron Levie on the impact of AI on jobs as told to FastForward this week.