Dental AI companies raised $800M in 2025. Where did it go?

Forty-three dental AI startups closed funding last year. $800 million poured into the sector. Not one changed how most practices actually work.

The companies are real. Dentech raised $80M for treatment planning AI. PatientFlow closed $65M for scheduling. DentAssist hit unicorn status with patient communication tech. But here's the gap: most of these products save your office 4 to 8 hours per week. Four to eight hours. On staff that costs you $30K/year, that's maybe $2.3K in annual savings. The software costs $300 to $600/month.

Investors loved the TAM story. 200K dentists in America. $200B industry. If AI takes 2 percent, that's $4 billion in new software revenue. The problem? Dentistry doesn't adopt. You've heard this about EHRs, practice management software, cloud backup. Adoption is slow. Resistance is structural. Change costs chairs in production.

What actually happened to that $800M? Marketing. Sales team. Runway. Investor meetings. Legal.

Your reality: Most of those companies won't exist in 2028. The ones that do will be acquired by DSO platforms or sold down 60 percent to existing vendors. You're not getting the miracle AI receptionist that cuts overhead. You're getting a $400/month feature inside your PMC that does 60 percent of what was promised.

OPERATOR MATH

Let's calculate the real ROI of AI software for a typical 3-doctor practice spending $500/month on a "revolutionary" AI patient communication tool.

Vendor promise: "Save 8 hours/week of staff time on patient communication." Annual cost: $500/month × 12 = $6,000/year. Promised savings: 8 hours/week × 48 weeks = 384 hours/year. Staff hourly cost: $18/hour (front desk). Promised annual savings: 384 hours × $18 = $6,912.

Actual results (based on typical adoption): First 90 days: Implementation, training, integration issues. Staff still manually handling 70% of communications because AI doesn't understand your PMS workflow. Effective time saved: 2 hours/week × 12 weeks = 24 hours. Months 4-12: AI is working better but staff still manually reviews everything. Effective time saved: 4 hours/week × 36 weeks = 144 hours. Total year 1 time saved: 168 hours = $3,024 in staff cost savings.

ROI calculation: Cost: $6,000. Savings: $3,024. Net loss: $2,976 in year one. Year 2 (if fully adopted): Time saved scales to 6 hours/week = 288 hours/year = $5,184 savings. Still not break-even. Meanwhile, your vendor raised $65M and spent $40M on sales and marketing, $15M on payroll, $8M on cloud infrastructure, and $2M on actual product development.

What actually saves time: A $300 virtual assistant (Upwork, Philippines-based) handling patient calls and texts for 20 hours/week = $6,000/year. Time saved: 20 hours/week × 48 weeks = 960 hours = $17,280 in equivalent staff cost. Net savings: $17,280 - $6,000 = $11,280/year. ROI: 188%.

THE TAKEAWAY

Before buying AI software: Ask for a 90-day pilot with actual usage data. Require proof: Show me practices like mine (size, specialty, PMS) using this successfully. What's their measured time savings? Can I talk to them? Calculate the staff time cost you're trying to save. If the software costs more than 70% of the staff time it saves, skip it.

What works (proven ROI): AI no-show prediction (if integrated with your PMS and you act on predictions). Automated insurance verification (if it actually verifies and doesn't require manual follow-up). AI-assisted radiograph analysis (if it catches things you'd miss and doesn't create false positives). Everything else? Mostly vaporware wrapped in VC funding.

The skeptic's playbook: When an AI vendor pitches you, ask: Who's actually using this in production for 12+ months? What's the measured ROI? Who's your biggest competitor and why should I pick you instead? If they can't answer confidently, they're selling a concept, not a product. Walk away. Your $6,000/year is better spent on proven tools (a virtual assistant, a better PMS, staff training) than unproven AI experiments.