Money is not the scarce resource in neurotech right now. Inside BCI reports that the sector closed the first quarter of 2026 with more than $960 million raised, and MedTech Dive says Synchron’s $200 million Series D in November 2025 pushed BCI investment across 2024 and 2025 past $1 billion. The market-size slides disagree wildly. One 2026 market report puts the global BCI market at $3.75 billion, while the Tech for Impact Summit’s 2026 overview says $8 to 12 billion. When the top-line numbers wobble like that, the diligence has to come from somewhere else.
This checklist is for the person about to write a first check, whether that is an angel investment, a syndicate slot, or a corporate venture pilot. Each of the seven points ends with things you can ask for in writing. It is a reading guide, not investment advice, and a securities lawyer should read the term sheet before you sign it.
Points 1 and 2: pin down the lane and the stage
1. Name the market lane first
Neurotech is three businesses under one label 🧭. A medical BCI answers to the FDA and to payers. A consumer neurowearable answers to marketing law and privacy statutes. A defense program answers to a contracting officer, and the GAO’s BCI overview notes that warfighters might one day operate a drone hands-free. Blur the lanes and a lab demo turns into a valuation. Ask the founder where the first dollar of revenue comes from, then check that the team, timeline, and burn rate match that lane.
Medical: revenue waits on an IDE, a registration-grade trial, marketing authorization, and a payer decision. The leaders raise nine-figure rounds to get there.
Consumer wellness: revenue can start early, but under the FDA’s revised General Wellness policy of January 6, 2026, disease claims push a product out of the low-risk lane.
Defense and government: grants and contracts pay early, and the customer, not the patient, writes the requirements.
Hybrid pitches: find the sentence where the deck switches lanes, because that is where the risk hides.
2. Verify the regulatory stage at the source
Regulatory language is where sloppy writing becomes a bad check 🔍. The Tech for Impact overview above calls Synchron FDA-cleared. It is not. Per Tech Times, Synchron holds an IDE and Breakthrough Device designation, and full approval waits on a 2026 trial meant to support the first PMA filing for an implantable BCI. Its Stentrode has 16 electrodes and sits in a blood vessel.
Precision Neuroscience shows the opposite trap. Its Layer 7 array (1,024 electrodes) holds a genuine 510(k) clearance, documented in FDA letter K242618. That clearance covers recording, monitoring, and stimulation for up to 30 days. The chronic wireless BCI is still in development. Breakthrough designation is a scheduling perk, not a verdict: Neurotech Reports notes that a designated device still faces 510(k), de Novo, or PMA review. Our ranking of neurotech breakthroughs by stage shows how far apart “cleared,” “in trials,” and “in a lab” sit. Pull up the last deck you received and circle every “cleared” and “approved” ⚖️. How many survive contact with the primary document?
The paper trail: the 510(k), De Novo, PMA, or IDE number, plus the exact indications for use.
Trial IDs: a ClinicalTrials.gov number for every study the deck mentions, with current enrollment status.
FDA meetings: minutes from the most recent Pre-Submission meeting.
Definitions: the company’s written meaning of “cleared,” “approved,” and “authorized” in its own materials.
Points 3 and 4: read the evidence and the hardware like an engineer
3. Read the trial like a skeptic
Synchron’s COMMAND study is a good early result that people still misread 📈. Clinical Trials Arena reports an early feasibility study in six patients with severe upper-limb paralysis, and all six met the primary endpoint of no device-related serious adverse events over 12 months. That is a safety result in a small group. It says the implant is tolerable. It does not say the implant works well enough to pay for.
Sample and endpoint: six patients answers a safety question, and efficacy needs a larger study with a pre-specified endpoint.
Peer review: a June 2026 ScienceDirect review says no formal peer-reviewed results from Neuralink’s feasibility studies exist as of early 2026, only announcements.
Controls: a 2025 meta-analysis of consumer neurofeedback finds typical samples of 30 to 50, mostly mindfulness-app controls, and a small pooled effect (g = -0.16).
Scoring: ask who funded each study and whether a blinded reviewer scored the outcome.
Cross-checking this by hand eats weeks. NeurotechMag Pro members can track every paper, funding round, and trial connected to developments like this one, summarized and searchable.
4. Interrogate the hardware and the IP together
Electrode count is a spec, not a performance claim 🧠. Sixteen electrodes in a vein and 1,024 on the cortical surface are both legitimate designs with different trade-offs between invasiveness and signal fidelity. What matters is what the decoder does at home, for months. A 2023 speech-BCI study in Advanced Science by Luo and colleagues reports stable decoding in a participant with ALS for three months without recalibration, and that is the kind of number to demand.
The IP question rides on the same hardware. Patsnap’s 2026 patent analysis advises mapping freedom to operate across three wireless power methods (RF inductive coupling, piezoelectric ultrasound, and magnetoelectric backscatter) and flags fast-growing university filings in China and South Korea 🛠️. A patent count tells you little. Claims that cover the way the product enters its market tell you more.
Home performance: decoding accuracy and output rate outside the lab, with weekly hours of use.
Recalibration burden: how often the decoder retrains, and who does it.
Longevity: the share of channels still usable at 6, 12, and 24 months.
IP opinion: a freedom-to-operate letter from outside patent counsel, plus the license terms for any university technology.
Points 5 and 6: follow the payer and the data
5. Map who pays after approval
Approval opens the door, and reimbursement decides whether anyone walks through it 💰. A PMC analysis of lessons from deep brain stimulation says implantable BCIs remain investigational in the US, and Medicare covers the routine costs of qualifying trials but not the investigational device itself outside specific programs. The National Security Commission on Emerging Biotechnology adds that no active CMS coverage exists for implantable BCIs, so no clear payment structure follows approval, and it says that gap discourages investment.
China shows how thin the word “reimbursed” can be. Neurotech Reports says China’s national insurance body assigned a billing code to Neuracle’s NEO within 48 hours of its March 2026 clearance. Yet a June 2026 landscape analysis on arXiv says approved BCIs are not yet broadly reimbursed nationally. A code is not a payment. Ask yourself who writes the check after the FDA does.
CMS contact: whether the company has used the point of contact the GAO says CMS provides for BCI developers.
Endpoint mapping: trial endpoints tied to coverage criteria, in the way Health Affairs Forefront argues for objective, function-based eligibility.
Price assumptions: the assumed price per patient, and the source for it.
Consumer path: for wearables, the cash price, refund rate, and subscription retention.
6. Audit the neural data rights
Brain data is the asset regulators are chasing fastest 🔒. Colorado, California, Connecticut, and Montana have each written neural data into state law, but the rules differ. Per Davis Wright Tremaine, only Colorado and Connecticut require opt-in consent for sensitive data, while California offers an opt-out. The federal MIND Act only directs the FTC to study the issue, and Inside BCI says it had not moved past committee as of April 2026. Practice lags the law: the Neurorights Foundation found that 29 of 30 companies selling neurotech online have access to brain data with no meaningful limits, per KFF Health News. Our piece on who owns your brain data counts nine or more state laws with different definitions, which is not a stable compliance target 🧾.
Data map: what neural data the product collects, where it lives, and the retention window.
Deletion: whether a user can erase neural data, not only the account, and whether “de-identified” data can be re-linked.
Legal regime: whether HIPAA, state privacy statutes, or both govern the product.
Exit clause: what happens to the data in a sale or shutdown, a question our piece on brain data when a startup shuts down walks through.
Point 7: underwrite the failure case
7. Price the company’s death, and the patient’s
Every implant company owes years of support after the last trial visit ⏳. Second Sight is the cautionary case. The Foundation Fighting Blindness reports that it invested more than $300 million over 24 years, earned under $32 million from its Argus II retinal implant over seven years, and shrank from more than 120 employees to 12 by May 2020. IEEE Spectrum documents the fallout for the 350-plus people who carry its implants 🏥. The GAO report above adds that some BCI trial participants had devices removed because no funds or medical support followed the trial. An investor who skips this point is quietly assuming the patient absorbs the loss 🧯.
Runway: months of cash to the next binary event, such as a trial readout or an FDA decision, not to the next board meeting.
Support reserve: a funded reserve or escrow for post-trial support and explant, and the name of whoever pays if the company dissolves.
Cap table: the preference stack, because a later, larger round can shrink an early check fast.
Plan B: what the company does if its lead indication fails, and who would buy the technology.
Run all seven on a startup you already like, then on one you don’t. If the deck survives both, you have a real conversation to start. Which of the seven does your favorite founder answer worst, and does the founder know it?