FrontierWhat is becoming possible?
The BCI Frontier
A map of what brain-computer interfaces can do, arranged by how firmly each capability is established: demonstrated in people at the centre, emerging possibilities at the rim. Choose a domain to see what is possible today, the latest milestone, who is working on it and what has to happen next.
- Demonstrated today
- Early clinical
- Research frontier
- Emerging possibility
Zones are qualitative editorial judgements. A capability sits in a zone, not at a point, and the map implies no ranking between domains.
Frontier domain
Motor Control
What is possible today
Participants have reached and grasped with robotic arms, controlled individual finger groups, and — through a brain–spine interface — stood and walked.
Latest important milestone20 Jan 2025
Continuous control of three finger groups gave one participant four degrees of freedom, used to fly a virtual quadcopter.
Capabilities by zone
- Robotic arm reach and graspDemonstrated today
- Walking through a brain–spine interfaceEarly clinical
- Multi-finger controlEarly clinical
- Exoskeleton control from epidural recordingsEarly clinical
- Dexterous whole-hand controlResearch frontier
- Natural-speed movement of the person's own limbsEmerging possibility
- Approaches
- Intracortical decoding of intended kinematics · Epidural ECoG driving spinal stimulation · Long-term ECoG control of robotic arms
- Clinical maturity
- Individual participants in academic feasibility studies. Brain–spine interfaces are moving toward company-led development.
- Labs
- University of PittsburghEPFL / Lausanne University HospitalClinatec, CEA / CHU Grenoble AlpesStanford UniversityUniversity of California, San Francisco
- Key papers
- Walking naturally after spinal cord injury using a brain–spine interfaceA high-performance brain–computer interface for finger decoding and quadcopter game control in an individual with paralysisReach and grasp by people with tetraplegia using a neurally controlled robotic armAn exoskeleton controlled by an epidural wireless brain–machine interface in a tetraplegic patient: a proof-of-concept demonstrationSampling representational plasticity of simple imagined movements across days enables long-term neuroprosthetic control
- Open datasets
- FALCON benchmarkAJILE12Handwriting BCI dataset
- On the calendar
- BCI Meeting 2027
- Patent themes
- 980 patent families concern motor decoding. Filing shows where protection is sought, not that a capability works. The patent landscape
What needs to happen next
- 01More participants in brain–spine interface studies
- 02Effectors robust enough for daily use outside the lab
- 03Combining movement with sensory feedback
Movement restoration is splitting into two paths: controlling external devices, and re-engaging the body's own circuits. The second has shown signs of promoting recovery, which may make it the more clinically compelling.
Unresolved and emerging
Open Frontiers
The questions that decide what happens next. Each is read across six dimensions, so that scientific momentum, clinical need and capital can be seen side by side.
Open question
Scientific momentum
Clinical need
Human evidence
Technical maturity
Competition density
Capital intensity
- Early human evidence
- High technical uncertainty
Published human data extend to years for a few participants and designs. Each new electrode technology has to earn its own record.
Start fromPlug-and-play control of a brain–computer interface through neural map stabilizationSampling representational plasticity of simple imagined movements across days enables long-term neuroprosthetic controlIntracortical
- Scientific momentum
- moderate
- Clinical need
- high
- Human evidence
- moderate
- Technical maturity
- low
- Competition density
- moderate
- Capital intensity
- high
Bars show low, moderate or high on each dimension. These are BCI Briefing judgements intended to make the landscape legible. They do not rank opportunities and are not investment advice.
The outlook
What happens next?
Four horizons, from now to beyond five years. Observed trajectories are on the left; plausible milestones, necessary breakthroughs and key uncertainties follow to the right.
Now
Observed trajectory
Speech BCI word accuracy over large vocabularies rose from about 76% to 97.5% in peer-reviewed reports between 2023 and 2024.
A high-performance speech neuroprosthesisAn Accurate and Rapidly Calibrating Speech Neuroprosthesis
Observed trajectory
Commercial implanted BCIs are in registered feasibility studies in five countries.
Observed trajectory
Six rounds of $200M or more appear in our sample of announcements since mid-2025.
Key uncertainty
Most company performance figures have not been independently measured or peer reviewed.
1–2 years
Plausible next milestone
A pivotal trial of a permanently implanted BCI is registered.
Plausible next milestone
Peer-reviewed results appear from commercial early feasibility studies.
Necessary breakthrough
High-performance decoders that stay calibrated for months without supervised retraining.
Key uncertainty
Whether signal quality from newer electrode designs holds beyond the first years in people.
3–5 years
Plausible next milestone
A first marketing authorisation for an implanted communication or control BCI in a major market.
Necessary breakthrough
Fully implanted, wireless, high-channel systems with multi-year reliability data.
Key uncertainty
Reimbursement, surgical capacity and long-term device support — the factors that decided the pace of cochlear implant adoption.
Beyond
Plausible next milestone
Bidirectional systems that combine movement and touch in everyday use.
Necessary breakthrough
A non-implanted signal with bandwidth approaching implanted systems. There is no human evidence for this yet.
Key uncertainty
How society chooses to govern interfaces for people without a medical need.
Only the first column describes what has happened. Everything to its right is a forward-looking judgement by The BCI Briefing as of 3 Oct 2026, and should be read as uncertain.