A product strategist, whose methodology is used by 80+ companies from San Francisco to Singapore. Trained 11,000+ product managers and worked with teams ranging from early-stage startups to companies with 100M+ DAU. As a CPO, he helped a tech company take ~25% market share from Google. An active mentor at Founder Institute, StartX (Stanford), Antler.
https://www.linkedin.com/in/ivan-zamesin/
Is the Golden Age of SaaS Ending?
Last month, a few founders I know canceled most of their SaaS subscriptions and didn’t buy new ones. Not because of budget cuts, but because they spent some time with Cursor and Claude Code. After just a few days, they had a custom knowledge base, a task tracker, and a CRM that worked better for their teams.
I keep hearing variations of this story from early-stage founders, marketers, product managers, and CTOs at companies of all sizes and industries. They’re not switching to a competitor—they’re building their own tools. And that might be a sign of something big happening in the SaaS market.
Here’s my hypothesis: the golden age of SaaS is coming to an end. Not with a dramatic collapse, but with a slow squeeze as vibe coding shifts the build-vs-buy equation that the entire industry was built on.
The Math That Made SaaS Work Just Broke
For many years, everyone understood the deal: building software is expensive and maintaining it is a pain. So you pay someone else to handle both.
But what happens when building software is no longer as expensive as it was just a few years ago?
The global SaaS market [hit $370 billion in 2025 and is projected to reach $842 billion by 2030](https://www.mordorintelligence.com/industry-reports/software-as-a-service-market). That’s a lot of subscription revenue. Those projections assume that building internally remains painful enough to justify paying for SaaS tools to do the job.
What if vibe coding—and the “vibe everything” approach—breaks that assumption?
When writing a PRD, a spec, and working code become five to ten times faster—and when people who aren’t engineers can actually participate—the “build it ourselves” option starts to look much more attractive for tools that once seemed like obvious subscription purchases.
I’ve talked to founders actively searching for product-market fit, and they all said the same thing: “Oh, we vibe-coded all our internal tools. Why would we pay for that?” It was spoken like the obvious move.
Of course, that’s not a survey—it’s anecdotal. But anecdotes at the edges often tell you where the middle is heading.
What’s Actually at Risk
Not every SaaS faces the same threat. The vulnerability depends on a few specific factors.
High exposure — tools without real moats:
– Task trackers
– Basic CRMs for small teams
– Dashboards
– Knowledge bases and wikis
– Simple workflow automation
These products share a pattern: relatively simple data models, well-understood interface conventions, and—most importantly—easy data transfer. If your customers can export a CSV and rebuild the core functionality over a weekend, you’ve got a problem brewing.
Lower exposure — products with defensible depth:
– Figma. Years of R&D in the collaborative engine, strong network effects, and design systems that lock teams in.
– Amplitude and Mixpanel. Data infrastructure, historical analytics that can’t be recreated, and all the necessary integrations.
– Salesforce. The ecosystem, the integrations, and the enterprise procurement relationships built over decades.
– Stripe. Payment infrastructure, compliance across dozens of countries, and all the stuff nobody wants to build themselves.
The dividing line is this: can a competent internal team, armed with AI coding tools, build a solution that fits their specific context better than a SaaS product designed for everyone—at a fraction of the yearly subscription cost?
The Unit Economics Squeeze
Let’s trace this logic to its uncomfortable conclusion.
Building software is getting easier and cheaper. More companies are building internally instead of buying. The ones still buying have more options, because new SaaS entrants also benefit from lower development costs. Competition intensifies. Customer acquisition costs rise.
Overall in the market, CAC has increased 60% over five years due to increased competition and ad spend in acquisition channels. That was before the current AI tooling wave. The trend hasn’t reversed.
And here’s the chain reaction that worries me: cheaper software development leads to more competitors, more competitors drive up acquisition costs, and rising CAC breaks unit economics in segments where it used to work fine. Eventually, it becomes economically unviable to compete for certain customer segments.
Now add internal competition on top of external competition.
For SaaS companies targeting SMBs and the mid-market with relatively straightforward tools, the unit economics might stop working. Not everywhere. Not overnight. But in cost-sensitive segments—where the product is easy to replicate, the data isn’t deeply locked in, switching costs stay low, and the customer has some internal technical capability—the economics get shaky.
What This Changes for Builders
If you’re building SaaS right now, the strategic questions look different than they did two years ago.
Question one: Can a motivated team with tools like Cursor and Claude Code rebuild your main functionality in a few weeks? If the answer is yes, then you might consider changing your market or focusing hard on things your clients can’t replicate in-house: unique algorithms and technology, strong network effects, integrations that are hard to copy, and compliance infrastructure that takes years to build.
Question three: Who are you building for? The vibe coding shift hits technical and semi-technical teams first. Traditional enterprises with no internal development culture move slower. Your target segment matters more than it used to.
Where I Might Be Wrong
Maybe the “build it ourselves” wave stays limited to early-stage startups with strong incetive to save costs. Maybe maintaining internal tools proves heavier than people expect.
But stories of people casually mentioning that they “vibe-coded their whole stack” are now common enough that it’s worth asking the question seriously: what happens to the SaaS market when a meaningful portion of potential customers decides they’d rather build than buy?
The golden age of SaaS rested on a simple truth. Renting was cheaper than building. That truth is starting to expire. And the companies that don’t see it coming will wonder why their pipeline dried up — not realizing they’re competing against their own customers now.
