AI companies will ultimately become biotech firms fighting to read and write your brain
Huberman predicts Meta, OpenAI, Anthropic, and Musk's Neuralink will all become biotech companies—the real arms race is non-invasive brain reading and writing, not AI models.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Mimicking Musk's four-hour sleep schedule is self-deception
Many want to copy Musk sleeping just four or five hours. Huberman says this is self-deception. There's no universal standard, but most people know their own number: for him, seven hours feels best, eight makes him drowsy, six is manageable, four or five hours for one night is fine, but two consecutive nights of insufficient sleep causes total ‘collapse’. Real self-assessment isn't mimicking high performers' sleep duration—it's honestly answering ‘At my current stage, how much sleep do I actually need to maintain peak performance?’
— Andrew HubermanPre-sleep downregulation requires completing three things in sequence
Patrick admits often needing four hours to shift from excitement into sleep. Huberman gives the sequence: first shift attention from thinking to pure sensation (listen to your heartbeat, breathing, room sounds), then trigger ‘respiratory sinus arrhythmia’ with extended exhales, further lower heart rate through the vagus nerve, finally make the body forget its limbs' positions—the method is closing your eyes and rolling them, mimicking the compensatory eye movements infants make when rocked. An MIT team is building an eyemask using electrical stimulation of muscles behind the ear to achieve the same effect, expected to launch in 7 to 12 months, and it can also increase REM sleep.
— Andrew HubermanAI giants' true endgame is becoming biotech companies
Huberman publicly predicts that Meta, OpenAI, Anthropic, and Musk's Neuralink will ultimately become biotech companies, not the AI-model companies outsiders assume. His reasoning: these companies' founders deeply understand neuroscience privately, and what they truly want is the ability to non-invasively read from and write to the human brain—such as precisely raising or lowering activation in specific areas like the nucleus accumbens to control motivation and focus. He calls this ‘the true arms race’, more fundamental than LLMs themselves.
— Andrew HubermanNeural firing sequence may be fundamentally irrelevant to memory
An experiment that unsettled neuroscience: chemically tag neurons active while animals learn a behavior, then later reactivate them in the original sequence. The animals precisely repeat the behavior—expected. But the control experiment found that reactivating the same neurons in completely reversed sequence, or like hitting all piano keys at once (activating them all simultaneously), produces identical behavior. This suggests the textbook assumption ‘neurons must fire in specific temporal sequence to form memory and action’ may be wrong. For technologies writing signals into the brain, this both lowers engineering difficulty and reveals massive gaps in existing brain models.
— Andrew HubermanInject light-sensitive ion channels into neurons for remote control
Huberman sketches the ‘cyborg’ path: inject a viral vector (not live virus—a gene-delivery vehicle) carrying a light- or ultrasound-sensitive ion channel gene into specific neurons. Afterward, wear a hat or glasses emitting long-wave light or ultrasound to selectively activate those neurons, adjust intensity, while AI automatically determines ‘what brain state this activation means for you’. He stresses the need for safety valves like those on vagal stimulators—otherwise people will abuse it to ‘cheat’ themselves.
— Andrew HubermanCRISPR babies' follow-up results are something no one dares ask about
He Jiankui (贺建奎) publicly announced at a conference that he had edited twin babies with CRISPR, deleting HIV receptors, claiming it would protect the children of an HIV-positive father. The scientific community's response was split: people who knew him privately asked ‘how's the progress,’ nobody publicly supported it until China's government shut down his lab and punished him. Huberman says he allegedly now operates a lab in Austin, Texas, but nobody can verify this. Less openly discussed: the HIV receptor gene also relates to neural circuits that form memories. Outsiders have long suspected he was simultaneously trying to make these children smarter.
— Andrew HubermanFour tech leaders are becoming neuroscientists, not just engineers
Huberman calls Sam Altman, Mark Zuckerberg, Dario Amodei, and Musk ‘the four athletes of brain exploration’—engineers who built platforms are now becoming neuroscientists, understanding neuroscience far deeper than outsiders realize. He says the biggest unsolved question in his field isn't sleep (well-researched) but awake brain states. We don't even have language to describe ‘focused yet relaxed conversation’ as a state. Only after clearly defining these awake brain states can we precisely induce them through technology.
— Andrew HubermanA mentor's willingness to let go is the kindest thing he's heard
During grad school, Huberman joined a trendy lab but secretly slipped away nights to another lab he truly cared about—but one that was unfashionable. Mentor Barbara Chapman discovered this but didn't demand he ‘return’. Instead she said he clearly loved that work more, that success never depends on whether the field is hot, only on how badly you want to know the answer. After transferring, he published eight first-author papers. He calls this the kindest thing anyone's done for him: it required his mentor to set aside ego, and it forced him to find the courage to say ‘I don't want to chase what everyone else is chasing’.
— Andrew HubermanIn their own words · checked verbatim
The Oracle said it best, like know thyself. You really have to be honest with yourself about what you need in order to do something for a long period of time.
Andrew Huberman6:14
anyone who thinks that Meta, OpenAI, Anthropic, or Elon's got Neuralink as his stated neural interest, anyone that thinks that those are LLM AI companies is perhaps going to be surprised at this prediction. Every single one of those is going to be a biotech company.
Andrew Huberman22:46
Turns out you get the exact same behavior. This is problematic for the field of neuroscience, because what it says is that this notion of the circuits firing in a particular sequence, at least for memory, it's probably not true.
Andrew Huberman35:33
It's not the active virus. The virus is just a vector. It's just a carrier for some genetic cargo that puts into your brain a channel such that if you have a hat on, wool cap or a baseball cap, that it can direct stimulation of only those neurons
Andrew Huberman43:53
He announced to the world at a meeting that he had done CRISPR in twin babies. He had deleted the HIV receptor.
Andrew Huberman53:10
And Sam Zuck and Dario, that's a pretty killer lineup. And Elon, those four, those are the big four. Those are the athletes of brain exploration now.
Andrew Huberman1:01:23
Success is determined by how precisely you match your genuine curiosity to the work that you do.
Andrew Huberman1:07:38
Figures
| Percentage of global adults consuming caffeinated beverages daily | 90%+ | 45:58 |
| Time to universal full-body MRI availability | 20 months | 47:00 |
| Percentage of entrepreneurs willing to try implantable vagal stimulation in 12 months | 20% | 30:00 |
Glossary
- Interoception
- The ability to sense one's own heartbeat, breathing, and other internal physiological states.
- Respiratory sinus arrhythmia
- The physiological phenomenon of heart rate decreasing during exhalation via vagus nerve activation.
- Proprioception
- The sense of one's own limbs' position and movement in space.
- Vagal stimulation
- Therapy using stimulation of the vagus nerve to regulate alertness or emotional state.
- Transcranial magnetic stimulation
- Non-invasive technique using magnetic fields to stimulate specific brain regions.
How to listen
Founders and investors tracking brain-machine interfaces and neurotech investment, plus high-intensity workers wanting to optimize sleep and focus.
17:57-21:46 criticizes the American health influencer ecosystem—many names but limited substantive insight, skippable.