Paradromics’ BCI: The Race to Restore Speech

The first human implantation of Paradromics’ BCI marks more than another milestone in brain-computer interface development. It highlights a different vision for the field, one centered on restoring communication rather than controlling a cursor or robotic arms.

On June 17th, surgeons at the University of Michigan Health implanted Paradromics’ Connexus device into the cortex of a woman living with motor neuron disease as part of the company’s Connect-One early feasibility study. The patient will be followed for six years under an FDA investigational device exemption that Paradromics says was the first IDE for speech restoration with a fully implantable BCI.

While many companies have focused on motor control, Paradromics is pursuing one of neurology’s most significant unmet needs: helping patients regain their ability to communicate.

Why Paradromics’ BCI Is Targeting Speech

The potential patient population extends well beyond a single neurological condition.

According to the World Health Organization, one in four adults will experience a stroke during their lifetime. In the United States alone, approximately 795,000 strokes occur each year, with ischemic strokes accounting for about 87% of cases. Around 30% of ischemic stroke patients develop aphasia, while nearly 47% experience some form of speech or language disorder. Together, that represents roughly 200,000 new aphasia cases annually.

Speech loss also affects most people living with ALS. Approximately 80% to 95% eventually lose the ability to speak. With roughly 6,700 new ALS diagnoses each year in the United States, more than 5,000 people annually are expected to lose their voice because of the disease.

Spinal cord injury represents another substantial population. Approximately 906 people per million are affected, equating to roughly 300,000 Americans, and as many as 87% experience speech disorders.

Today, speech and language therapy remains the primary standard of care. While therapy can improve communication, results vary considerably between patients, and progressive diseases such as ALS continue to advance despite treatment.

Paradromics is attempting to address that gap through implantable neural technology.

What Makes Paradromics’ BCI Different?

Connexus is a fully implantable, high-density microelectrode array designed to capture neural activity directly from the cortex. The coin-sized implant contains hundreds of electrodes connected to a chest-implanted transceiver, which wirelessly sends signals to an external receiver. Software then converts those signals into text, synthesized speech, or device control.

The company’s strategy is built around neural bandwidth. Decoding speech requires substantially more information than identifying movement direction, making high-resolution recordings and large channel counts central to Paradromics’ approach.

University of Michigan researchers have already demonstrated that the device can be implanted, record neural activity, and be removed intact in less than 20 minutes. That procedural efficiency could become especially important for patients who may require future revisions.

Since its founding in 2015, Paradromics has maintained comparatively modest funding. Compass AI records show an $18.3 million DARPA award in 2017, Seed financings in 2018 and 2021, a $33 million Series A led by Prime Movers Lab in 2023, and an investment from NEOM Investment Fund in 2025. Reporting places the company’s venture funding above $105 million, alongside approximately $18 million in NIH and DARPA grants.

A Crowded BCI Landscape

Paradromics is competing against several substantially larger players in the BCI market.

Neuralink completed a $650 million Series E financing in 2025 at a reported $9 billion valuation and had implanted more than a dozen patients by late 2025. Synchron, backed by Jeff Bezos and Bill Gates, raised approximately $200 million while advancing its endovascular Stentrode platform. Precision Neuroscience received FDA clearance in 2025 for its temporary thin-film cortical array, becoming the first next-generation wireless BCI to achieve a commercial foothold. Blackrock Neurotech, according to Compass AI, has raised roughly $210 million over its history.

Despite being smaller, Paradromics has chosen a distinctly different regulatory pathway. Rather than focusing first on motor restoration, the company is pursuing speech restoration, earning the first investigational device exemption specifically for a fully implantable speech restoration BCI.

That decision may ultimately prove strategically significant.

Communication presents a more straightforward clinical endpoint than broader concepts such as neural control. Either a patient regains meaningful communication through the device, or they do not. That clarity has implications for clinical studies, regulatory review, reimbursement, and future commercialization.

Compass AI’s Predictions

Compass AI’s Prediction Agent (a Beta capability) expects several important developments.

First, it predicts Paradromics will report its first in-human speech or communication decoding results from the Connect-One study before the end of 2027, most likely at a neuro or BCI conference.

Second, Compass AI expects a major medtech strategic, potentially a company such as Medtronic, to acquire an ownership stake in or establish a partnership with an implantable BCI company before the end of 2027 as the category continues maturing.

Finally, Compass AI predicts that the first implantable BCI to enter a pivotal or registrational pathway will target speech or communication rather than motor prosthetics, reflecting the combination of a larger unmet need, clearer clinical endpoints, and a stronger reimbursement story.

Although Paradromics remains in the early stages of clinical evaluation, Paradromics’ BCI has positioned itself around one of the field’s clearest unmet needs. Whether that strategy ultimately proves successful remains to be seen, but the race to restore speech may become the most important competition in the next generation of brain-computer interfaces.