I work adjacent to the DFW venture capital scene and what I am seeing with AI startups reminds me of every bubble I have studied. But there are important differences.
Signs we ARE in a bubble:
-
Valuations are disconnected from revenue. DFW AI startups with $500K in ARR are raising at $50M+ valuations. That is a 100x multiple. For context, healthy SaaS companies trade at 10-15x revenue.
-
Every startup is an "AI company" now. I have seen pitch decks where the AI component is literally a ChatGPT API call with a custom prompt. That is not an AI company. That is a wrapper.
-
Hiring is irrational. DFW companies are paying $300K+ for "AI engineers" whose primary skill is writing Python scripts that call OpenAI's API. When the correction comes, these salaries will normalize.
-
The "picks and shovels" companies are printing money. NVIDIA's valuation, cloud GPU providers, AI tooling companies — this mirrors the dot-com era where Cisco and Sun Microsystems peaked before the crash.
Signs this is NOT the dot-com bubble:
-
The technology actually works. Dot-com companies were selling vaporware. AI companies are selling products that measurably improve productivity. The ROI is real.
-
Revenue exists. OpenAI reportedly hit $5B+ in annualized revenue. Anthropic is growing rapidly. These are real businesses with real customers.
-
Enterprise adoption is genuine. Every Fortune 500 company in DFW is spending real budget on AI integration. This is not hype-driven — it is CFO-approved.
-
The infrastructure investment has lasting value. Data centers being built in DFW (there are 3 major ones under construction in the metroplex) will have value regardless of which AI companies survive.
My prediction:
- 80% of current AI startups will fail or get acqui-hired. Normal startup mortality.
- The underlying technology is real and transformative.
- We will see a correction in AI valuations within 12-18 months, followed by a more rational growth phase.
- DFW is well-positioned because our AI economy is enterprise-focused, not consumer-hype-focused.
Sources:
- PitchBook — DFW AI startup funding data
- NVIDIA quarterly earnings
- Personal conversations with DFW-based VCs and founders
Bubble or not, the smart money is building on the technology while being realistic about timelines.
The wrapper problem is real. I have evaluated 15 AI startups for my company this year. At least 10 were just UIs on top of OpenAI with no defensible technology. When OpenAI adds that feature natively, those startups evaporate.