TL;DR
The "38% of B2B buyers built internal AI tools instead of buying SaaS" statistic is a viral social-media exaggeration with no backing from credible research. But the underlying threat is real and narrower than the headline suggests: early-stage startups and tech teams are "vibe-coding" lightweight custom alternatives to mid-tier SaaS using LLMs and platforms like Replit, forcing founders to compete on defensibility, enterprise-grade security, and integrations — not just AI features. For enterprises, build-vs-buy remains heavily weighted toward buy, but only if you offer what custom tooling can't: compliance, maintenance, and seamless integration into sprawling legacy stacks.
The Viral Statistic That Broke: Separating Signal from Noise
The "38% of B2B buyers built an internal tool with AI instead of buying SaaS" claim is a social-media conflation, not a finding from credible analyst research.
The statistic originated from viral Instagram and LinkedIn posts by venture capitalist Ash Rust (Sterling Road) in May 2026, but there is no survey data from Gartner, Forrester, McKinsey, or G2 confirming that 38% of B2B buyers have built internal software replacements. Instead, the "38%" figure is highly likely a conflation of three separate G2 and Demand Gen statistics: 38% of B2B buyers use AI agents to shortlist vendors, 38% trust generative AI platforms to assess solutions, and 38% prefer engaging with sales reps over individual research.
This matters because it reframes the actual threat. The real competitive pressure isn't that enterprise buyers are replacing SaaS wholesale — it's that a specific segment (early-stage startups and tech-forward teams) is building lightweight, niche alternatives to mid-tier software. The enterprise buyer's calculus remains fundamentally different: security, compliance, and maintenance overhead still favor buying over building.
What to watch: Whether the "vibe-coding" trend expands beyond startups into mid-market companies, or remains confined to small teams with technical depth and low regulatory burden.
The Real Threat: Lightweight Custom Tooling at Startup Scale
The actual build-vs-buy shift is happening at the margins, not at enterprise scale, and it's driven by a specific economic arbitrage.
Ash Rust details in The Founder's Playbook how his firm and Y Combinator startups are using LLMs and development platforms to build custom alternatives: instead of paying $30,000/year for a premium VC CRM like Affinity, teams built custom, tailored CRMs using a $100/year Replit plan. For simple use cases — basic data connections, workflow automation, lightweight reporting — the cost-to-capability ratio has inverted. A non-technical founder can now prompt Claude or GPT-4 to generate a functional tool in hours rather than weeks.
The implication is that your real competitor is "good enough" custom tooling: if your product only offers basic workflows and thin wrappers around AI APIs, enterprise buyers will increasingly opt to build a lightweight internal prototype instead. This doesn't mean you're losing to a competitor — you're losing to the buyer's own engineering team. The only antidote is defensibility: deep integrations, enterprise-grade security, collaboration features, and a user experience that cannot be easily replicated by a non-technical employee prompting an LLM.
What to watch: Whether your product's core value sits in workflow simplicity (vulnerable to custom builds) or in integration depth and compliance (defensible against build).
Why Enterprises Still Buy: The Build Illusion at Scale
Enterprise buyers face a different equation entirely, and most underestimate the true cost of building.
This is why enterprises are consolidating rather than fragmenting: rather than fully replacing SaaS with custom builds, enterprise buyers are using AI to build highly targeted, lightweight internal tools for niche workflows while consolidating their core software stack around platform vendors. The math is clear — the maintenance burden and integration complexity make building a full-featured system prohibitive at scale.
The strategic implication is that your vulnerability to build isn't at the enterprise level — it's at the startup and mid-market level, where technical depth and regulatory burden are lower. If you're selling to enterprises, your risk isn't displacement by custom tooling. Your risk is being displaced by a platform vendor that offers a superset of your functionality and integrates more deeply into the buyer's existing stack.
What to watch: Whether your enterprise customers are consolidating around platform vendors or maintaining best-of-breed point solutions — if consolidation is accelerating, your long-term defensibility depends on being either the platform or deeply embedded in one.
The Visibility Moat: Where Custom Tooling Can't Compete
The most defensible advantage against build-vs-buy pressure isn't product features — it's visibility in the AI-generated vendor shortlist.
Because 38% of buyers are using AI agents to shortlist vendors, your online footprint matters more than ever. If your product is not actively recommended in AI search results and comparison queries, you may never even make the initial consideration set. A startup team considering whether to buy or build will first prompt an LLM: "What are the best CRMs for venture capital?" If you're not in that answer, the build decision is already made.
This is why brand authority, analyst positioning, and review platform presence have become prerequisites for survival. Custom tooling can replicate your features, but it can't replicate your brand visibility in LLM-generated comparisons. The founder selling to enterprises must ensure they're actively recommended by AI tools before any human conversation begins.
What to watch: Whether your product appears consistently in LLM-generated vendor comparisons for your category — if it doesn't, you're competing against both custom builds and visibility blindness simultaneously.
What surprised us
-
The 38% statistic is pure social-media conflation. There is no credible research backing the claim that 38% of B2B buyers have built internal AI tools instead of buying SaaS. The number is a viral post by a VC that conflated three separate G2 statistics about buyer behavior. This is a reminder that the loudest voices in tech often amplify noise, not signal.
-
The real build-vs-buy pressure is at startup scale, not enterprise. Enterprises are consolidating around platforms and recognizing that maintenance costs make building prohibitive. The threat to mid-tier SaaS is from startups and small teams that can "vibe-code" lightweight alternatives for $100/year instead of $30k/year. If you're selling to enterprises, your risk isn't build — it's platform consolidation.
-
Defensibility now requires compliance and integration depth, not just features. The only way to compete against custom tooling is to offer what a non-technical team can't build in a weekend: SOC 2 compliance, seamless integrations into legacy stacks, and maintenance burden transfer. Feature parity is no longer enough.