Source: Unsplash
On May 7, 2026, Neuralink announced its next-generation surgical robot can now place electrode threads into virtually any region of the human brain. The engineering community collectively lost its mind. But as someone who builds interfaces for a living, I had a different reaction: who is actually designing what the user experiences once that chip is in?
Brain-computer interfaces (BCIs) are entering a new era in 2026, one where the hardware is scaling fast, the surgery is getting cheaper, and implants are moving from restoring lost motor function to potentially treating Parkinson's, epilepsy, and depression. Neuralink's robot once cost between $10 million and $20 million to manufacture. That figure is now around $500,000. More than 20 patients across the US, UK, Canada, and UAE have already received the N1 implant. The market is projected to grow from $3.33 billion in 2026 to $13.86 billion by 2035. The hardware story is real, verifiable, and accelerating. The design story is barely a draft.
"Neuralink will start high-volume production of brain-computer interface devices and move to a streamlined, almost entirely automated surgical procedure in 2026. Device threads will go through the dura, without the need to remove it. This is a big deal."
— Elon Musk, X (formerly Twitter), January 2026
It is a big deal. I am not dismissing that. But when the founder of a medical device company calls something "a big deal," they are almost always talking about the engineering. The robot. The electrode density. The surgical precision. The signal fidelity. And those things genuinely matter. Brain surgery that once required removing part of the skull is now automated, faster, and dramatically cheaper. That is progress worth celebrating.
What is not being celebrated, or even discussed seriously, is the interface layer. The thing the user actually touches. Or in this case, thinks at. Because right now, if you have a Neuralink implant, your primary interaction model is still basically a computer cursor. You think about moving your hand. The chip reads the motor cortex signal. A cursor moves on screen. You play chess. You browse the web. You type at maybe 40 words per minute if you are Brad Smith, the ALS patient who became Neuralink's third implant recipient earlier this year.
That is genuinely life-changing for someone with ALS. I am not minimizing it. But it is also, from a UX standpoint, a glorified mouse. And if BCIs are going to move beyond accessibility tools for people with paralysis into something broader, the interface design work needs to start now. Not after the hardware is "good enough." Now.
The Hardware Milestone Is Real. The Design Gap Is Bigger.
Here is what the May 2026 announcement actually means. Neuralink's new robot can navigate to any region of the brain, not just the motor cortex. Previous generations were limited in reach, which is why early use cases were all about restoring movement. The new system uses advanced cameras and sensors to compensate in real time for the microscopic brain movement caused by heartbeat and respiration. It threads electrodes directly through the dura mater. The precision is genuinely remarkable.
This opens the clinical door to applications far beyond motor restoration. Parkinson's disease. Refractory epilepsy. Treatment-resistant depression. Neuralink describes the long-term goal as a "generalised neural interface" covering neurological conditions across the brain. That is a fundamentally different product than what exists today. And it needs a fundamentally different design approach.
Think about what that means from an interface perspective. A motor cortex implant has one job: read movement intention, translate it to a cursor. The mental model is simple. But a generalised neural interface touching regions responsible for mood, memory, language, and perception? That is a product with dozens of overlapping interaction modes, each with different latency tolerances, error costs, and user mental states. The design complexity explodes. And nobody is seriously working on it yet.
Five UX Problems BCI Teams Are Not Solving
I have spent the last several years designing interfaces for complex enterprise systems, and there are clear parallels to what BCI teams are walking into unprepared. Here are the problems I would be losing sleep over if I were a product designer at Neuralink or Synchron:
- Error tolerance and recovery flows: Brain signals are noisy. Even a 96% accuracy rate sounds impressive until you apply it to consequential actions like sending a message, confirming a transaction, or dismissing a medical alert. Touchscreens have undo. Voice interfaces have "cancel." What is the undo mechanism when your brain misfires and the system interprets it as a confirm? This design problem does not exist yet in any public BCI roadmap I have seen.
- Mental fatigue and session design: Using your motor cortex to move a cursor is exhausting. Early Neuralink users reported significant cognitive load during extended sessions. If BCI interfaces are going to be used for hours at a time, the session design matters enormously. When does the system prompt rest? How does it detect fatigue? What does a "log out" even mean when the device is always on?
- Consent and intent disambiguation: How does the operating system distinguish between a conscious command and an idle thought? If I am thinking about clicking something, versus actually intending to click it, the interface needs a model for intent confidence. This is not just a technical problem. It is a design philosophy problem that needs to be solved at the interaction layer.
- Trust signaling: Right now, users get limited feedback about what the system "heard" from their neural signals. A well-designed BCI interface needs to show users their own interpreted commands before execution, like a preview layer that builds trust and reduces anxiety. Nobody is building this yet in any meaningful way.
- Accessibility at scale and the class problem: Synchron's Stentrode is implanted through the jugular vein, far less invasive than Neuralink's cranial surgery. It now connects with NVIDIA AI processing and Apple Vision Pro, letting paralyzed patients control digital environments through thought. But the fundamental trade-off remains: less invasive means lower signal quality, meaning a performance gap built into the hardware that no amount of UX design can fully overcome. This creates a tiered BCI experience, premium invasive hardware for those who can access and afford it, lower-resolution non-invasive options for everyone else.
The most dangerous assumption in the BCI industry right now is that better hardware will eventually make the interface problems go away. It will not. The interface problems will scale with the hardware. They will just become more consequential.
Synchron vs. Neuralink: Two Different UX Bets
It is worth looking at the two leading approaches here because they represent very different product philosophies, and both have serious design implications.
Neuralink is betting on high-fidelity signal capture. More electrodes, closer to neurons, in more brain regions. The interface quality ceiling is higher, but the surgical barrier is also higher. The product assumption is: if the signal is rich enough, we can build any interface on top of it. It is an engineer's bet.
Synchron is betting on accessibility. The Stentrode is inserted via the jugular vein. No brain surgery. No craniotomy. The recovery is measured in days, not weeks. The signal quality is lower, but the risk profile is dramatically better. And interestingly, Synchron has moved faster on the integration side. The Stentrode already connects with Apple Vision Pro and NVIDIA's AI processing stack. Paralyzed patients are controlling 3D environments through thought. That is a more complete product story, even if the underlying hardware resolution is lower.
From a product designer's perspective, Synchron is doing something smart. They are building partnerships with existing interface ecosystems. They are not trying to design an entirely new interaction model from scratch. They are plugging into frameworks that users already understand, Apple's spatial interface, NVIDIA's AI processing, and extending them with neural input. That is the more realistic near-term path to usable product design.
I actually wrote about the risk of building entirely new interaction models from scratch in a piece on Medium earlier this year. The companies that win in emerging interface paradigms are rarely the ones who try to reinvent everything. They are the ones who find the right bridge between what users already know and what the new hardware makes possible. Synchron gets this. Neuralink, at least publicly, does not seem to yet.
The Security Problem Nobody Is Talking About Enough
On May 16, 2026, TechTimes reported that researchers have now formally documented and simulated a family of BCI cyberattacks. Neural Flooding. Neural Jamming. Neural Spoofing. Neural Sinkhole. These are not theoretical. They are documented attack vectors that exploit the wireless communication layers of invasive BCIs to disrupt or induce neuronal activity.
Let that land. Researchers have now mapped out how to attack a device that is literally threaded into your brain.
The privacy situation is equally alarming. A report by the Neuro-rights Foundation found that more than 90% of consumer neuro-technology companies use vague safeguarding language with no concrete protection of consumers' neural data. And as of 2026, no federal law specifically governs BCI cybersecurity standards. Your neural data, the electrical activity of your literal thoughts, has less regulatory protection than your credit card number.
ISACA's 2026 blog called this "the CISO's dilemma: when your employees' thoughts become attack vectors." That is not science fiction. That is the enterprise security problem product teams need to be designing around right now. Consent interfaces, data minimization by default, clear visualisations of what neural data is being transmitted and when: these are design requirements, not engineering afterthoughts.
What Needs to Happen Before Any of This Scales
I am not pessimistic about BCIs. The progress is real and the potential is extraordinary. But I have watched enough emerging tech categories to know how this usually goes. The hardware scales. The software catches up eventually. The design layer gets bolted on last, after the product is already in users' hands, and the user experience suffers for years because of decisions made early that were never revisited.
Here is what I think needs to happen, from a product and design perspective, before BCI interfaces can scale beyond clinical trials:
First, intent modeling needs to become a first-class design component. The interface needs a concept of "confidence in intent" built into its visual language. If the system is 95% sure you meant to do something, that should look different than 65% sure. The user needs to be part of the feedback loop, not just a signal source.
Second, BCI companies need to hire product designers, not just neuroscientists. The teams building these products are overwhelmingly engineers and researchers. That is appropriate for the current phase. But the moment these devices start targeting conditions like depression and epilepsy, the users are no longer just people with motor impairments. They are people with complex cognitive states, emotional variability, and entirely different mental models of what a "brain interface" should do. You need people who know how to design for that complexity.
Third, the industry needs open interface standards before it is too late. Right now, Neuralink and Synchron are building proprietary stacks. That is fine for now. But as the market grows to $13.86 billion by 2035, the lack of interface standards will fragment the ecosystem in ways that hurt users. Think about the mess that early mobile app ecosystems were before iOS and Android established baseline interaction patterns. BCI needs that conversation now, not in 2030.
The Designer's Real Role Here
I have been designing AI-native products for long enough to know that the interface layer is always where the product either earns trust or loses it. It does not matter how good the underlying model is. It does not matter how precise the electrodes are. If the user cannot build a coherent mental model of what the system is doing with their neural signals, they will not trust it. And without trust, the technology never scales past niche clinical use cases.
Brain-computer interfaces are not a hardware problem anymore. The hardware is already extraordinary. They are now a design and trust problem. And the clock is ticking. Because once high-volume production starts, once the surgical robots get cheaper, once the regulatory approvals start coming through, the interface decisions made in the next 18 months will define the user experience of BCI for a decade.
We explored the broader challenge of designing trust into AI systems at reloadux.com/blog earlier this year, and the same principles apply here, maybe even more urgently. The stakes are higher. The interface is more intimate. And the consequences of getting it wrong are not just a bad user experience. They are a bad experience that is surgically implanted in someone's skull.
What do you think? Are BCI companies moving too fast without the design foundations in place, or is it still too early to worry about the interface layer? Drop your thoughts in the comments below. I genuinely want to know what people building and using these systems are experiencing on the ground.
Sources:
1. Neuralink Surgical Robot Reaches Any Brain Region, May 2026 — https://www.techtimes.com/articles/316912/20260520/neuralink-robot-reaches-any-brain-region-parkinsons-epilepsy-depression-now-within-surgical.htm
2. Neuralink's Next Robot Could Reach Any Part of the Brain — https://yourstory.com/2026/05/neuralink-surgical-robot-brain-interface-2026
3. Brain Computer Interface Market Size, Precedence Research — https://www.precedenceresearch.com/brain-computer-interface-market
4. Ghost in the Shell: Researchers Document BCI Cyberattacks, May 2026 — https://www.techtimes.com/articles/316738/20260516/ghost-shell-july-7-neuralink-scales-researchers-document-bci-cyberattacks.htm
5. Brain-Computer Interface 2026: Neuralink, Synchron, and Real Progress — https://3zebras.com/science/brain-computer-interface-2026-neuralink-synchron-update/15397/
6. ISACA: The CISO's Dilemma When Employees' Thoughts Become Attack Vectors — https://www.isaca.org/resources/news-and-trends/isaca-now-blog/2026/the-cisos-dilemma-when-your-employees-thoughts-become-attack-vectors
7. Elon Musk on Neuralink High-Volume Production — https://www.fiercebiotech.com/medtech/elon-musks-neuralink-kickstart-high-volume-production-brain-computer-interface-devices