Since ChatGPT exploded into the mainstream, it’s been blamed (or credited) for rewiring everything it touches, from killing homework shortcuts to reshaping hiring. But now AI is spooking the very industry that built it: software.
And the market is reacting like it just saw a ghost.
While “big tech” keeps levitating, a lot of “medium tech” software names are getting smoked. Former SaaS royalty like Salesforce $CRM, Adobe $ADBE, and Atlassian $TEAM now trade at revenue multiples under ~5x sales. Meanwhile the iShares Expanded Tech Software ETF $IGV is down more than 7.5% YTD.
This isn’t just a normal rotation. It’s a narrative reset.
Software just got de-moated
The core fear: AI-native “agentic” tools are making software cheaper, faster, and easier to build.
Tools like Anthropic’s Claude Code have become the poster child, but OpenAI’s ChatGPT and Alphabet’s Gemini are pushing the same message: a lot of what used to justify SaaS pricing power is getting automated away.
Investors are now asking a brutal question:
If AI can build workflows, write code, generate dashboards, and stitch together automations… what exactly is the moat anymore?
Vibe-coding startups are not a meme, they’re traction
This panic isn’t theoretical. It’s being reinforced by real adoption curves.
Per Similarweb traffic data, “vibe-coding” platforms have seen monthly traffic surge over the past year as more people experiment with building apps from prompts instead of programming from scratch.
Lovable is the headline example: it reportedly went from a $1M revenue run rate to $100M in just eight months. Its CEO even described the mission as “building the last piece of software.”
Another platform, Emergent, reportedly just tripled its valuation after reporting rapid growth.
Public markets don’t like this story at all, because it implies “software = less scarce.”
The public market hit is now spilling into private markets
Public software stocks getting repriced is one thing, it’s visible, painful, but clean.
Private markets are messier.
Because venture valuations are less transparent and liquidity is delayed, this is where the slow-motion damage spreads. A huge chunk of venture’s past decade is built on one assumption: SaaS exits at high multiples.
Now those exit multiples are collapsing in real time.
Harry Stebbings summarized the VC anxiety pretty clearly: the venture model struggles when public revenue multiples shrink this hard.
VC money is still in software, but it’s moving inside software
Software has been venture capital’s favorite parking spot for decades.
PitchBook data shows software consistently took ~25% of US VC dollars in the 2010s. That dominance got even bigger recently: in 2025 alone, software startups absorbed roughly $172B, about 53% of all VC invested.
But here’s the real change: the money inside the sector is rotating.
Last year, “AI and machine learning” startups pulled a bigger share of VC funding than traditional SaaS companies for the first time.
So while SaaS is getting boxed in by public market repricing, it’s also getting squeezed in private markets because capital is leaving for AI-native startups.
Chamath’s take: the SaaS meltdown is here
Chamath Palihapitiya went full blunt mode and called this the start of a “Great SaaS Meltdown.”
His point is simple:
SaaS has historically sold investors on “grow now, profit later.”
But if AI makes software creation and workflows dramatically cheaper, that long-term harvest may never arrive the way it was modeled.
In other words, the growth might not be durable, and the profits might not show up.
That’s the double hit.
The math is ugly: SaaS was priced for a world that may not exist anymore
This is the part that makes VCs sweat.
For years, SaaS companies raised capital at double-digit revenue multiples because software was viewed as the ultimate sticky, scalable asset.
PitchBook data shows SaaS valuations averaged well above 10x revenue in the 2010s, then averaged roughly ~22x from 2020 to 2025. Over that stretch, the sector absorbed an estimated ~$466B in venture capital.
But now public comps are closer to ~4x to 5x sales.
That changes everything.
Because if public SaaS exits aren’t paying venture-style multiples anymore, then a lot of late-stage private valuations were basically built on sand.
Bottom line: software fear might be overdone… but it’s not baseless
It’s still unclear how much of this is a real permanent reset versus the market overreacting to the newest tech shock.
The skepticism is fair too: is a dentist in Idaho really going to vibe-code their own scheduling software?
Probably not.
But Wall Street isn’t pricing “dentists will code.” It’s pricing something more subtle and more dangerous for incumbents:
Software is getting easier to recreate.
And when scarcity disappears, multiples usually follow it out the door.