Neuralink N1 vs Blackrock Utah Array: New Flagship vs Twenty-Year Standard
One has 10× the channels and the other has 20 years of evidence. The headline-grabbing implant vs the workhorse most landmark BCI papers actually used.
Neuralink's N1 is a 1,024-channel intracortical implant: wireless, inductively charged, surgically inserted by a custom robot, and currently the highest-channel-count clinical-stage human BCI. Blackrock's Utah Array is a 96-electrode silicon array that has been implanted in dozens of patients since 2004, holds an FDA 510(k) clearance (NeuroPort) since 2011, and is the hardware behind essentially every landmark BrainGate result — handwriting, speech, robotic-arm control. N1 has more channels. Utah has more papers.
Published · Updated
Side-by-side specs
| Specification | N1 Implant | Utah Array |
|---|---|---|
| Classification | ||
| Invasiveness | invasive | invasive |
| Primary modality | single-unit | single-unit |
| Direction | read | read-write |
| Electrodes | ||
| Total channels | 1024 | 96 |
| Recording channels | 1024 | 96 |
| Electrode type | penetrating-shank | penetrating-shank |
| Prep time | — | — |
| Acquisition | ||
| Sampling rate | 18600–19300 Hz | — |
| ADC resolution | 10 bit | — |
| Connectivity | ||
| Protocols | bluetooth-le, transcranial-inductive | — |
| Power | ||
| Battery life (active) | — | — |
| Physical | ||
| Weight | — | — |
| Software | ||
| Raw data access | No | Yes |
| LSL support | — | — |
| SDK | ||
| Has SDK | No | — |
| Open source | No | — |
| Regulatory | ||
| FDA status | investigational-device-exemption | 510k |
| CE mark | No | — |
| Pricing | ||
| MSRP | — | — |
| Subscription required | — | — |
| Warranty | — | — |
Verdict by axis
channel-count
N1 Implant
1,024 vs 96 — N1 has roughly 10× the channels of a single Utah array.
Confidence: high
signal-quality
tie
Both record single-unit activity. Utah's 30 kHz acquisition (via Cerebus) edges N1's 18-19 kHz, but at the level of spike sorting both are research-grade. N1's flexible threads may track moving cortex better; Utah's rigid silicon is the proven long-term substrate.
Confidence: medium
invasiveness
tie
Both require craniectomy. The percutaneous pedestal Utah needs is a long-term infection vector that N1's wireless design avoids.
Confidence: high
research-credibility
Utah Array
The Utah Array is the single most-cited intracortical BCI device in the literature. Hochberg 2006, Willett 2021, Willett & Kunz 2023 and dozens of others all run on this hardware. Neuralink has not yet published equivalent peer-reviewed clinical results.
Confidence: high
clinical-evidence
Utah Array
Two decades of human-implant data, three Nature papers in five years, dozens of patients. N1 has six months of single-patient progress reports.
Confidence: high
fda-status
Utah Array
Utah Array (as the NeuroPort System) holds an FDA 510(k) clearance from 2011. N1 is investigational under an IDE.
Confidence: high
availability
Utah Array
Utah Arrays are sold to qualified researchers via Blackrock's institutional channels. N1 is investigational only.
Confidence: high
stimulation
Utah Array
Utah Array supports intracortical microstimulation when fitted with iridium-oxide tips, enabling sensory feedback experiments. N1 is read-only as currently studied.
Confidence: high
capabilities-cursor
tie
Both have demonstrated 2D cursor + click control in humans. The bandwidth ceiling on N1 is higher in principle but not yet shown in published clinical data.
Confidence: high
capabilities-typing
Utah Array
Stanford / BrainGate's high-WPM handwriting decoder runs on Utah arrays (Willett 2021). Equivalent results on N1 have not been published.
Confidence: high
capabilities-prosthetic
Utah Array
Multi-DOF robotic-arm control across multiple patients has been published using Utah arrays. Not yet shown on N1.
Confidence: high
ecosystem
Utah Array
Cerebus + Central + Open Ephys + the BrainGate consortium = a deep, well-known stack. Neuralink's tooling is internal and proprietary.
Confidence: high
durability
Utah Array
Utah arrays have multi-year human implant histories; signal quality does degrade due to gliosis, but the failure modes are characterised. N1's first patient experienced significant thread retraction within months of implant.
Confidence: medium
developer-experience
Utah Array
Utah arrays plug into a rich research stack (Cerebus, Central, BrainFlow, Open Ephys). N1 has no public SDK.
Confidence: high
Pros & cons
N1 Implant
In favor
- 1,024 recording channels — highest in any clinical-stage human BCI
- Fully wireless (BLE + inductive charging) — no transcutaneous connector
- Custom on-implant ASIC with spike detection
- Robot-assisted insertion targets surgical reproducibility
- Flexible polyimide threads may reduce micromotion-induced gliosis
Against
- Investigational only (IDE) — not commercially available
- No peer-reviewed clinical trial paper as of 2026-05
- First-patient thread retraction event raised durability questions
- Most hardware specs proprietary
- Read-only (no microstimulation)
Utah Array
In favor
- FDA 510(k) cleared since 2011 (NeuroPort)
- Two decades of human-implant track record across dozens of patients
- Bidirectional (recording + microstimulation with IrOx tips)
- The hardware behind every landmark BrainGate result
- Commercially available to qualified researchers
- Mature ecosystem (Cerebus, Central, BrainGate) with deep documentation
Against
- Only 96 channels per array — substantially fewer than newer devices
- Requires open craniectomy + a percutaneous pedestal connector (infection vector)
- Wired — no wireless option
- Signal quality degrades over months to years due to gliosis
- MR-unsafe in most clinical contexts
Recommendations by use case
| Use case | Pick | Why |
|---|---|---|
| Cursor + click decoding research today | Utah Array | Available, peer-reviewed, FDA-cleared, with a large existing patient cohort. |
| High-WPM typing / handwriting decoding | Utah Array | Willett 2021 set the bar on Utah arrays. N1 has not yet matched it in publication. |
| Speech decoding from motor cortex | Utah Array | Willett & Kunz 2023 demonstrated 62-WPM continuous speech on Utah arrays. |
| Robotic-arm / multi-DOF prosthetic control | Utah Array | BrainGate has published this on Utah; N1 has not. |
| Sensory feedback / bidirectional BCI | Utah Array | Utah arrays with IrOx tips support intracortical microstimulation. N1 is read-only. |
| Maximum future bandwidth | N1 Implant | 1,024 channels is more than 10× a single Utah array — the headroom is there if the rest catches up. |
| Fully wireless, no skin-penetrating connector | N1 Implant | N1 is wireless. Utah Array uses a percutaneous pedestal. |
| Long-term implant stability research | Utah Array | Multi-year Utah recordings exist; N1's longest implant is just over a year. |
| Buying one for personal use | Neither | Both are investigational devices in clinical trials. Neither is purchasable. |
Frequently asked
›Why does Neuralink get the press if Utah is the more proven device?
Public attention. Utah has been quietly powering cortical BCI research for two decades; most of the field's biggest results — the BrainGate cursor demos, the Stanford handwriting decoder, the 2023 speech BCI — were obtained on Utah arrays. Neuralink is more visible, but as of 2026-05, the peer-reviewed clinical evidence base behind Utah is dramatically larger.
›Can I buy a Utah Array?
Only if you are a qualified researcher with an institutional account and the appropriate IRB / ACUC / IACUC protocols. Blackrock sells the array (and the Cerebus acquisition system) through institutional channels, not retail.
›Why are flexible threads better than rigid silicon?
Theoretically, flexible electrodes track brain motion (cardiac pulsation, breathing, head motion) without tearing tissue, which should reduce gliosis and improve long-term recording stability. The tradeoff is that they're harder to insert — Neuralink's R1 robot exists specifically to solve this. Whether the theoretical advantage holds up over years of human implantation is still unknown.
›Does N1 have stimulation?
Not in the current PRIME study. The N1 implant is read-only as currently configured. Utah arrays equipped with iridium-oxide tips support intracortical microstimulation and have been used for sensory-feedback experiments.
›How many channels do I actually need?
Depends on the paradigm. Cursor control with a click can work on 16 channels; high-WPM typing typically uses 96–192; the most ambitious speech-decoding work uses 200–400. Neither device is the limiting factor for the most common BCI applications today — both have enough channels for cursor-class tasks.
Bottom line
For research today, the Utah Array is the right answer — it's available, FDA-cleared, supports stimulation, and underpins essentially every landmark BCI paper. N1 is the more ambitious device (10× the channels, fully wireless) and may eventually surpass Utah on every axis, but as of 2026-05 the published clinical evidence isn't there yet.