Is AI actually killing SaaS companies and software jobs? According to the revenue data — not yet, and not even close. Despite a wave of alarming headlines and a brutal stock selloff (some SaaS names are down 50%+ year-to-date), the actual revenue numbers from public cloud software companies are still solidly in growth territory. That doesn't mean the threat isn't real. It just means the SaaS apocalypse, for now, is a stock market story — not a business fundamentals story.
Two CEOs laid out their starkly opposing worldviews in the same weekend edition of the Wall Street Journal: Mark Benioff of Salesforce, who thinks the bears are completely wrong, and Verizon's new CEO Dan Schulman, who is predicting 20 to 30 percent unemployment within five years. One is fighting for survival in the AI era. The other is trying to grab headlines. Both are worth understanding.
Is AI Actually Killing SaaS Companies Right Now?
The short answer: no — at least not in the revenue data. The narrative that AI agents will replace SaaS subscriptions and hollow out the software industry has caused massive stock selloffs. Salesforce is down 28% year-to-date. Some competitors are down nearly double that. But when you look at what's actually happening to revenue, the picture is completely different.
Here's a snapshot of recent revenue growth across major public SaaS and cloud companies:
- Cloudflare: +34%
- Snowflake: +30%
- Data Dog: +29%
- Monday.com: +25%
- GitLab: +23%
- HubSpot: +20%
- Workday: +14.5%
- UiPath: +14%
- Adobe: +12%
- Salesforce: +10.8% (and reaccelerating)
- Box: +9%
- Asana: +9%
- Zoom: +5.3%
Take GitLab, for example — a company that, in theory, should be getting absolutely destroyed. It's open source. Developers can fork it and vibe code on top of it. They don't need to pay GitLab. And yet GitLab is growing at 23%. The apocalypse just isn't showing up in the numbers. The expectation reset is real. The actual business deterioration? Not yet.
What Is Salesforce Agent Force and Is It Working?
Salesforce launched Agent Force in 2024 — its flagship AI product designed to let businesses deploy AI agents that can autonomously handle real customer interactions. As of now, about 23,000 of Salesforce's 150,000 customers are using it. That's roughly 15% adoption, which is meaningful but far from widespread.
Early reviews were mixed. Customers complained that they had to spend nearly half their time just preparing and cleaning data so the AI could understand it — which seriously limits the platform's effectiveness right out of the box. Salesforce has since built a data integration layer into its tech stack that automatically pulls in customer information from external sources, and it's acquired several companies specializing in data management and AI-powered sales tools.
Where is it actually working? At Pearson, the education company, Agent Force now autonomously handles queries about order statuses, refunds, and lost access codes. The result: a 40% increase in customer questions resolved without any human interaction. That's a real, measurable outcome.
Where is it struggling? Complex, nuanced customer interactions — the kind that require human judgment. Pandora Jewelry's chief digital officer said Agent Force couldn't reliably recommend products based on vague, natural language inputs like "my wife likes dogs, what should I buy her?" That's a genuinely hard problem, and it highlights the gap between what AI can do in controlled demos versus real-world retail scenarios.
What Is Salesforce's Secret AI Platform Agent Albert?
Beyond Agent Force, Salesforce is building something bigger. Code-named Agent Albert — a nod to their existing Einstein AI branding (Albert Einstein, get it?) — this is described as a new AI platform that automatically studies its users and takes actions on their behalf. Benioff plans to unveil it by the end of 2025.
The project has been in development for three years. After ChatGPT launched and caught the entire industry flat-footed, Benioff convened a standing Saturday morning meeting to accelerate Salesforce's AI efforts. That urgency is now crystallizing into Agent Albert, which is positioned as the next generation beyond Agent Force — not just answering queries, but proactively acting on behalf of users across the entire Salesforce ecosystem.
Benioff's broader argument is that the leading AI labs — OpenAI, Anthropic (in which Salesforce is an early investor), Google — couldn't replicate what Salesforce offers even if they tried. The security, compliance, and deep CRM integrations that enterprise customers rely on aren't things you can vibe code overnight. "People think we have our back against the wall when in fact the opportunity has never been greater," he said.
Which SaaS Companies Are Still Growing Despite AI?
Almost all of them, actually. The growth rates listed above tell a consistent story: the market is pricing in a future disruption that hasn't materialized in the income statements yet. This is important context for investors, founders, and operators trying to understand whether to panic or plan.
Salesforce itself is an interesting case study in deceleration. Back in 2012, it was growing at 36%. By 2020, still a healthy 28.7%. Growth peaked and then started sliding — 18% in 2023, 11% in 2024, 8% in early 2025. But the most recent quarters are showing reacceleration: 8.7%, then 9.6%, then 10.8%. That's not a company in freefall. That's a mature enterprise software giant finding its footing in a new AI landscape.
Why Is the Verizon CEO Predicting Massive AI Job Losses?
Verizon's new CEO Dan Schulman has made waves by predicting 20 to 30 percent unemployment within two to five years — a number so extreme it would exceed Great Depression levels. He's also warned that humanoid robots will begin disrupting manual labor jobs previously considered safe, and he's pushed hard for reskilling and retraining programs.
To put that in context: even Dario Amodei of Anthropic — someone genuinely on the frontier of AI concern — only predicted 50% unemployment among entry-level white collar workers. But the US only has 5 to 7 million such workers in a 170-million-person labor force. If Dario's scenario came true in full, overall unemployment would land somewhere between 6 and 9%. That's rough, but it's not Great Depression territory. Schulman's 20 to 30% would require essentially no government intervention, no Fed action, and an impossibly fast simultaneous takeoff of both AGI and humanoid robotics.
There's also an obvious irony here: Verizon's 13,000 layoffs — the largest in company history — were explicitly described by Schulman as not related to AI. The company was, in his words, "too hierarchical, too bureaucratic, too process-oriented." It was a traditional corporate bloat reduction. And yet Schulman is simultaneously warning employees to prepare for AI to take their jobs, recommending they ask an AI to write their own obituary to understand the technology, and creating a $20 million career transition fund.
Is the 20 to 30% number a genuine prediction or a headline grab? It's worth considering that the biggest number always wins the news cycle. Either way, Verizon is a strange candidate for the AI disruption narrative — they own spectrum allocations and cell tower infrastructure. You can't vibe code a cell carrier.
Will AI Really Cause 20–30% Unemployment?
Probably not — and the current data supports that skepticism. AI-driven unemployment isn't showing up in US labor statistics. AI-driven SaaS churn isn't showing up in revenue data. The Philippines customer service sector — frequently cited as an AI apocalypse bellwether — accounts for roughly 6 to 7% of that country's economy, not the 90% some podcasters have claimed.
The more credible near-term scenario comes from the Boston Consulting Group: AI will reshape roughly half of US jobs in two to three years, with up to 15% potentially eliminated outright. Even that figure would be partially offset by new job creation — suggesting a transition period of perhaps 6 to 10% unemployment, not a civilization-ending collapse.
Jensen Huang, Andy Jassy, and most major tech CEOs remain bullish that AI creates more jobs than it destroys — just as every prior wave of automation did. The difference this time may be the speed of transition. But the deployment rate of even proven autonomous technology (like self-driving cars) remains well below 1% of total vehicles on the road. Fast takeoffs are much harder in the physical world than in the discourse.
What Happened With the Blue Origin New Glenn Launch Failure?
On a separate but notable front, Blue Origin suffered a setback on its third New Glenn rocket mission. The rocket launched successfully from Cape Canaveral, and — impressively — the massive booster returned and landed safely, a feat only Blue Origin and SpaceX have achieved with orbital-class rockets. But the payload, a satellite for AST SpaceMobile (which is building a cellular broadband network in space), was deployed into an incorrect orbit.
The satellite's altitude was too low to sustain operations and was subsequently deorbited. AST SpaceMobile said the loss would be covered by insurance, and the stock recovered from an initial 16% overnight drop to about a 6% decline by end of day. This type of mishap isn't unprecedented — SpaceX experienced a Falcon 9 upper stage failure in 2024 that sent Starlink satellites to the wrong orbit. Growing pains for an emerging commercial launch business. Blue Origin will be back.
The bottom line across all of this: the gap between the AI narrative and the AI reality remains enormous. SaaS companies are still growing. Mass unemployment hasn't arrived. Salesforce is building new tools and fighting back. And even a high-profile rocket mishap resolves faster than the doom-and-gloom headlines suggest. Stay grounded in the data — it's almost always less dramatic than the discourse.
