The Highest-Value AI Frontier Is Shifting from Interpretation to Acquisition

Most radiology AI analyzes images after a skilled operator captures them. Butterfly’s GA Tool automates the scan itself — eliminating the operator skill requirement entirely. That is a structural distinction, not a marketing one. If blind-sweep AI proves replicable, handheld ultrasound becomes a software-defined medical instrument where value accrues to the AI layer, not the probe.

Status Quo

Interpretation-Layer AI

1,400+

FDA-cleared radiology AI devices

Algorithms analyze images after a trained operator has already captured them. Competes for radiologist attention. Adoption gated by workflow integration.

Crowded

The Shift

Acquisition-Layer AI

1st

FDA-cleared blind-sweep ultrasound AI

AI automates the scan itself. No sonographer training required. Gestational age estimate in under 2 minutes. Attacks the scarcity constraint directly.

Category-creating

Butterfly Platform — Hover to explore each layer

        Clinical Pathway AI NEW
        GA Tool · FDA Cleared
      
First blind-sweep ultrasound AI. Estimates gestational age with zero sonographer training. Pipeline includes pre-eclampsia, malpresentation, and multiple gestation detection — same scan protocol.

        Acquisition-Assist AI
        Guidance · Auto-capture
      
Probe guidance, image quality scoring, and auto-capture. De-skills the scanning process at the point of acquisition — the foundation the GA Tool builds on.

        Enterprise Software
        Compass · Garden SDK
      
Cloud workflow platform with fleet management, billing integration, and analytics. Butterfly Garden SDK opens the platform to third-party AI developers.

        Hardware Platform
        iQ3 · ~$2,000
      
Single-probe, whole-body ultrasound at the lowest price point in the category. Butterfly Move cart system bridges handheld and traditional form factors. Octiv chip licensing extends the platform to OEM partners.

        Semiconductor Foundation
        Ultrasound-on-Chip
      
Proprietary Deep Silicon ASIC. Three hardware generations (iQ → iQ+ → iQ3) driven by Moore’s Law cost curve. Multi-generation IP with co-development partnerships.

↑ Value is migrating from hardware to software

$97.6M

FY2025 Revenue
(+19% YoY)

21M+

AI Training
Images

1.8M

Scans in
Sub-Saharan Africa

<2 min

GA Estimate
Time

What to Watch Next

Catalyst

Follow-on fetal condition AI — same blind-sweep protocol

Risk

No CPT code — reimbursement pathway undefined

Catalyst

US rural health deployment contracts + federal funding

Risk

Label scope: 16–37 weeks only; not a full anatomy scan

Catalyst

Octiv chip licensing to OEM partners

Risk

Cash consumption — not yet consistently profitable