Brain-computer interfaces moved from speculative to clinical in the past few years. The progress is real and small. The marketing, especially around consumer applications, remains decades ahead of the technology. This update separates the two.
We talked to clinicians and reviewed the published literature on what's actually happening in patients.
What changed in the past two years
The clinical side is finally a real category.
- Neuralink moved past first patients. Multiple participants now have implants and are doing public demonstrations of cursor control.
- Synchron expanded its trials. Their stent-based "Stentrode" now has FDA breakthrough designation and a growing patient cohort.
- Several non-invasive consumer EEG companies pivoted or shut down. The gap between EEG marketing and clinical performance kept widening.
How current BCIs actually work
Two architectures dominate the clinical category.
- Implanted electrode arrays. Neuralink uses a flexible thread array placed in the cortex. Higher resolution, surgical risk, longer setup.
- Vascular electrodes. Synchron's Stentrode is delivered through blood vessels — no open brain surgery. Lower resolution, much lower surgical risk.
Both record neural signals; both need machine learning models to translate intent into action. The actual user experience for patients today is using a cursor, typing through a custom interface, and operating simple devices.
Where it works and where it doesn't
Restoring basic communication and motor control for patients with severe paralysis (ALS, spinal injury) is the active clinical use case. Patients can type at slowly-improving rates and control basic cursors. For someone who couldn't communicate before, this is life-changing.
Beyond that — restoring vision, complex motor control, sensory input — the technology is at much earlier stages. Single-electrode demos exist; usable systems do not.
Consumer applications are not on a near-term horizon. The current invasive interfaces require neurosurgery, immune compatibility, and ongoing maintenance. None of that scales to healthy users.
Comparison: BCI approaches in April 2026
| Approach |
Maker |
Use case |
Status |
| Cortical electrode array |
Neuralink |
Cursor + comms |
Human trial |
| Vascular electrode |
Synchron |
Cursor + comms |
Human trial |
| ECoG (surface) |
Various academic |
Speech decoding |
Research |
| Consumer EEG |
Muse, Emotiv |
Meditation, gaming |
Limited utility |
Common misconceptions
"Neuralink will let you type with your thoughts in 5 years." Maybe for paralyzed users. For healthy people who can already use a keyboard, the risk-benefit doesn't justify it.
"EEG headsets read my thoughts." They read averaged brain electrical activity through the skull. The signal is too noisy and slow for fine control.
"Memory implants are next." Memory restoration trials exist (DARPA-funded work in epilepsy patients), but they target specific memory types in narrow conditions, not general memory enhancement.
What to actually expect this decade
The realistic 2026–2030 outlook:
- More patients enrolled in trials. Hundreds, not thousands.
- Improved typing and cursor speeds for paralyzed users.
- First sensory restoration demonstrations (vision, touch).
- No consumer-grade invasive BCI.
If this disappoints, the comparison point is right — pacemakers took decades from first prototype to common implant. BCIs are following a similar curve.
FAQ
Could I get a BCI today?
Only as a clinical trial participant with a qualifying condition. Outside trials, BCIs are not commercially available.
Are non-invasive BCIs useful?
For relaxation and very basic command interfaces, yes. For meaningful control or "reading thoughts," no.
Is Neuralink safe?
Early trials have proceeded without major reported incidents. Long-term safety, including immune response and electrode degradation, will only be known after years of patient follow-up.
Where to go next
For related guides see Apple Vision Pro in 2026: where the platform is two years in, Quantum computing in 2026: where the field is, and Smart glasses 2026: Meta vs Ray-Ban.