“Will AI Kill SaaS?” Is The Wrong Question

Last week saw a severe meltdown of software company stocks on Wall Street.

Salesforce: −47%
Figma: −83%
Hubspot: −71%
ServiceNow: −53%
Workday: −45%

Click here for a more complete list.

According to Wall Street Journal, the $300 billion drawdown in software stocks was caused by the threat posed by new AI vibe coding tools to them.

Investors’ fears that new developments in artificial intelligence will supplant software reverberated through the stock market Tuesday, dragging down the shares of companies that develop, license and even invest in code and systems. – Wall Street Journal.

Termed SaaSpocalypse, the meltdown spread to SaaS-adjacent firms like data providers, SaaS investors, and IT Services companies e.g. Expedia, Infosys, LegalZoom, Reuters, TCS.

We’ve been hearing that “AI will disrupt SaaS” for nearly a year now. I shared my thoughts from the technology pov in Will ChatGPT Kill SaaS? and other related posts over the last 6-12 months.

Looks like Wall Street took the AI threat to SAAS seriously last week.

I’ll cover the implications of that watershed moment in this post (Spoiler alert: “Will AI kill SaaS?” is the wrong question).

This is not the first rodeo for “A will kill B“.

We’ve heard this posturing many times in the 30 years since Late Professor Clayton M. Christensen of Harvard Business School published his theory of Disruptive Innovation in the mid ’90s:

PC will kill Mainframe
Client Server will kill Datacenter
eBook / Kindle will kill Print Book
EV / Tesla will kill ICE
Ecommerce / Amazon will kill Retail / Walmart
Rideshare / Uber will kill Yellow Cab
Fintech will kill Bank
Digital Payment will kill Cash
Crypto will kill Fiat.

As you can see, all these narratives take the form

Entrant will kill Incumbent.

For those who are not familiar with the disruptive innovation theory, incumbent is the leader in a certain technology and is typically a large and publicly traded company; and entrant is a scrappy new kid on the block with some disruptive innovation in the space dominated by the incumbent. I’ll spare the details of how disruption is posited to happen other than to note its definition: According to the original theory, disruption meant that the incumbent shuts down; this was later diluted to the end state in which the incumbent enters a terminal decline and sees its revenue whittle down to a fraction of its peak value i.e. to $50 million in 1996 dollars.

Fueled by the disruption posturing, entrants got high valuations, and incumbents lost market cap. Click here if you wish to know how disruption posturing helped drive valuations of startups.

Cue to the present day. IBM still sells mainframes. The hottest area of tech now is datacenters. Far from dying, traditional banks in USA posted more than a trillion dollars in profits in the last decade. I could go on and on but my drift should be clear: None of the aforementioned disruptions happened in actual practice.

Unless I’m missing something in this recent picture of New York Hilton Midtown, it doesn’t look like Uber / Lyft has disrupted yellow cabs in NYC. #Consensus2019 #CoinDesk #NewYorkHiltonMidtown pic.twitter.com/ottkqvstO5

— GTM360 (@GTM360) June 12, 2019

However, markets did not revert back to mean.

Incumbents never reclaimed their peak market caps even after the disruption posturing fell flat. In many cases, entrants continued to enjoy nosebleed valuations e.g. Tesla. The #1 EV company reported year-on-year decline in sales in its latest fiscal year. It withdrew its higher-end car models. Nobody believes any longer that EV will kill ICE automobiles. Despite all that, Tesla continues to have a higher market cap than any ICE automobile company. (Last I checked, Tesla was more valuable than the next 10 automobile companies put together.)

This is a nod to the famous quip by John Maynard Keynes, “markets can stay irrational for longer than the trader can stay solvent.”

I suspect we’ll see a repeat of this movie in AI v. SAAS.

Ironically, SaaS, which is now on the RHS of the disruption equation, began life on the LHS of it 30 years ago.

Until the mid-to-late 1990s, software meant Commercial Off The Shelf software. COTS was licensed and installed on the customer’s infrastructure, ergo it was also called onprem software. The top three onprem CRM software vendors at the time were Siebel, Oracle and SAP. Siebel was a pureplay CRM company whereas Oracle and SAP added CRM to their suite of onprem enterprise applications.

Then came Salesforce with a CRM software hosted on the cloud in what’s generally regarded as the first ever SaaS software.

The narrative that SaaS will disrupt onprem software took root and quickly spread like wild fire. Many SaaS companies entered the market. On the back of dotcom boom leading up to year 2000, they succeeded in raising massive venture capital at extremely high valuations. Simultaneously, Wall Street hammered down the stocks of onprem vendors. Unable to survive as an independent company, Siebel was acquired by Oracle. (although SAP and Oracle managed the onprem to SaaS shift just fine).

Enterprises did not switch to SaaS anywhere as fast as SaaS Disruptionistas predicted. In fact, it took 25 years for just half of enterprise workloads to cross over from onprem to SaaS.

tl;dr: Even 30 years after Salesforce entered the market, SaaS has not come anywhere close to disrupting onprem.

However, that didn’t stop SaaS companies from enjoying rising valuations and onprem vendors from suffering falling market caps during this period.

Now the shoe is on the other foot.

SaaS has become the incumbent. AI has become the entrant. “AI will disrupt SaaS” has become the consensus opinion.

Enterprises have started to rip off their SaaS and replace them with AI-built code that delivers the tiny fraction of SaaS functionality that they currently use. SaaS Repatriation, as I call this trend, is growing rapidly.

The market priced that in last week. OpenAI, Anthropic and other AI companies are raising capital at ever-rising valuations; Salesforce, ServiceNow, Workday and other SaaS companies are seeing a steady erosion in their market caps.

Some companies are doing AI BUILD inhouse whereas others are getting it done by third party IT Services companies.

They’re not. Their body shopping and fixed price IT Services vendors are going to AI BUILD the tiny subset of ERP, BM, CRM that they actually use. As had happened before COTS BUY started displacing BUILD status quo around 35 years ago.

— SKR (@s_ketharaman) February 5, 2026

While there’s a still long way to go for AI vibe coding to replace SaaS, let’s not forget that SaaS took over 25 years just to overtake onprem software.

Even if AI never kills SAAS in the end, many SaaS companies may die midway, and the ones that are left may never reclaim their peak valuations. (Not investing advice).

Wall Street is filled with communities that tend to run in herds. The past couple of weeks have definitely seen the herd collectively conclude that software is dead. – @stevesi @a16z.

AI might do to SaaS what SaaS did to onprem software.

Rinse and repeat.