BOM Cost Estimator

ESP32-S3

Basic AI (voice, gesture). WiFi + BLE built-in.

Rockchip RV1103

Camera + AI, 0.5 TOPS NPU. Plaud NotePin tier.

Rockchip RK3562

Mid-range AI, 1 TOPS NPU, Mali G52 GPU.

Rockchip RK3588S

High-end, 6 TOPS NPU, octa-core. For edge servers.

MediaTek Genio 510/700

4–10 TOPS NPU, industrial grade.

512MB + 4GB eMMC

Basic: voice recorder, simple sensor device.

1GB + 8GB eMMC

Standard: camera device, on-device AI inference.

2GB + 32GB eMMC

High: smart display, local LLM, multi-model AI.

8MB PSRAM + 4MB Flash

MCU-tier: ESP32-S3 builds, no discrete RAM.

No display

Audio-only device (like Plaud NotePin).

0.96" OLED (mono)

Simple status display, 128x64.

1.54" AMOLED (color)

Smartwatch-tier, 240x240.

2.4" TFT LCD IPS

Full color display, 320x240.

2.13" E-ink

Ultra-low power, 212x104.

100–150mAh

Ultra-slim pendant, ~4hr runtime.

300–500mAh

Compact wearable, ~8–12hr runtime.

800–1000mAh

Full-day device, ~16–20hr runtime.

1500–2000mAh

Multi-day, larger form factor required.

WiFi + BLE (built-in)

Included in ESP32-S3. No extra cost.

BLE 5.0 module

nRF52840-based, ultra-low power.

WiFi/BT module (add-on)

For non-Espressif SoCs.

4G LTE Cat-1

Quectel EC21/EC25. Adds SIM slot + antenna.

GPS/GNSS add-on

Location tracking. +$1.50–8 on top of above.

MEMS Microphones (×2)

PDM/I2S. Essential for voice AI.

6-axis IMU

Accelerometer + Gyroscope.

Camera 2MP (OV2640)

Basic AI vision, fixed focus.

Camera 5MP (OV5640)

Higher resolution, auto-focus.

Environmental (temp/humidity/pressure)

SHT30 + BMP280 combo.