I moved cities 9 times before college, so I stopped expecting friendships to last.
Every 3 years my father got a new posting, and every 3 years I started over.
I'd make friends, we'd promise to stay in touch, and within a few months we wouldn't.
So I learnt to keep things light.
I made friends knowing I'd leave them.
Then, in 6th standard in Hyderabad, I met a girl from a Navy family.
I moved away soon after, like always.
But she did something nobody else did.
She called me on my birthday that year, and the next year, and every single year after that.
→ Through four more cities and schools
→ Through college, Adobe and starting up
→ Through all the years I forgot to call back
She's still my oldest friend, and she still lives in Hyderabad.
I used to think long friendships happen to lucky people.
Now I know they happen because one person refuses to let go.
Tag the friend who never let you drift away.
Every year of my journey looked like a mistake to someone.
2015
Got AIR 190 in JEE Mains and picked IIIT Delhi over NIT Warangal.
2017
Became the only second-year intern on my team at Adobe.
2018
Turned down Adobe's full-time offer, my dream company.
2019
Presented a research paper in Greece and joined a fellowship instead of a job.
2020
Started a medical device company in the middle of a pandemic.
2021
Walked into AIIMS every day for three months wearing two masks.
2024
Got India's medical licence for EASE and presented at the World Health Assembly in Geneva.
2026
Forbes 30 Under 30 Asia and pitched in Silicon Valley.
None of these looked smart in the year I made them.
They only make sense now, in a line like this.
Onwards.
@narendramodi
I once said something out loud in a room that was worth over a billion dollars, and I had no idea how I'd do it.
It was my first event at the Plaksha Tech Leaders Fellowship in 2019.
I was 22, fresh out of IIIT Delhi, sitting with founders and investors whose combined net worth crossed a billion.
When it was my turn, I said, "I'm going to build a neurotech company, either right away or after my PhD."
I didn't have a product, a co-founder or a plan.
I had a sentence.
A year later, Lakshay and I incorporated Marbles Health.
We had met in that same fellowship.
Today, doctors across India use EASE to treat depression and anxiety, and we've crossed 26,000 brain stimulation sessions.
Here's what saying it out loud actually does:
→ People start sending you the right introductions
→ You stop treating the idea as a someday thing
→ You feel embarrassed enough to actually follow through
Most people wait until they're sure before they say it.
I said it first and figured out the rest later.
What's the sentence you haven't said out loud yet?
Unpopular opinion:
A lot of what we call overthinking is just a bad night's sleep.
After one night without sleep, your brain's alarm centre, the amygdala, reacts over 60% more strongly to upsetting things.
At the same time, the part of your brain that normally calms it down loses its grip.
So the email feels ruder, and the problem feels bigger. Nothing changed except your sleep.
"Sleep on it" isn't just old advice.
It's neuroscience.
No, you cannot sleep 5 hours, sit through 6 back-to-back calls, answer Slack in between, skip lunch, and still think clearly by 5 PM.
Microsoft actually tested this. They put EEG caps on people in back-to-back video calls and watched stress build in their brains with every call. When people took 10-minute meditation breaks between calls, it stayed flat.
Feeling fried by evening isn't laziness.
It's biology.
Free yourself from the guilt.
Plan A:
• Crack JEE
• Get into a top college
• Join Adobe
• Settle down by 25
Plan B:
• Turn down Adobe
• Build a brain stimulation device
• Get it medically licensed
• Take it to 75+ hospitals 100+ now
Meet the PM
• Explain it to every relative at every wedding every occasion
Plan B is going well.
The weddings, not so much.
Two of AI’s biggest unsolved problems are things the human brain handles while you sleep.
⠀
A new review in Neuron, co-authored by researchers at Google DeepMind, maps where today’s AI still falls short.
⠀
It can prove math theorems better than most people, yet it struggles with everyday tasks like booking a flight.
⠀
The authors trace that gap to 4 things the brain still does better:
1. It keeps learning without erasing old memories
2. It replays experiences during rest and sleep
3. Weak memories get replayed more, so they stick
4. It constantly checks how well it is doing
⠀
The first one remains unsolved for most AI systems.
When a model learns something new, it tends to overwrite what it already knew.
⠀
The brain solves this with the next two, and much of that work happens during sleep. It replays the day, gives extra attention to what was learned weakly, and locks it in without losing the rest.
⠀
Self-checking is the other gap. A model’s confidence is only loosely tied to whether it is right, which the authors link to hallucinations.
⠀
Rest is one of the brain’s most active states, and AI research keeps coming back to it for answers.
⠀
Source: “The recipe for intelligence in natural and artificial systems,” Neuron, 2026
Researchers strapped cameras to little children’s heads, and what they learned is changing how we train AI.
One famous AI vision model, CLIP, learned from 400 million images paired with text.
A baby learns to see with far less, and with blurry eyes.
Here is how a baby’s vision develops:
⠀
• It starts out very blurry at birth
• It sharpens slowly over the first months
• Faces and big shapes come in first
⠀
When researchers trained AI on head-camera footage from young children, the models still learned useful visual skills from a tiny slice of that data.
And when they trained AI blurry first, the way babies see, the models got better at recognising objects in new situations.
The order matters for people too.
Children born with cataracts that are removed later miss that early blurry stage, and some show lasting difficulty recognising faces.
I work on brain health every day, and I keep coming back to how much timing shapes the brain.
It turns out AI might learn better with something like a childhood.
Sources: Radford et al., ICML 2021 (CLIP); Orhan et al., NeurIPS 2020 (head-camera models); Cai et al., CVPR 2026 (blurry-first training); Le Grand et al., Nature 2001 (cataracts and face recognition)
Your brain speaks two languages at once, and medicine has mostly learned one of them.
Every time a brain cell sends a message, two things happen in a fraction of a second:
• An electrical pulse runs along the cell
• At the end, it releases a chemical messenger
• That chemical sparks the next cell’s pulse
So every thought, mood and memory you have runs on electricity and chemistry working together.
Medication is how medicine learned to work with the chemical part. It adjusts those chemical messengers, and for millions of people it genuinely helps.
But for around 60 to 70% of patients, medication only gets them part of the way, and that’s where the electrical side comes in.
Gentle brain stimulation works with that electrical side, helping specific brain areas get back into a healthy rhythm while medication keeps doing its job on the chemistry.
That is the idea behind EASE at Marbles Health, a second pillar of care that sits right next to medication.
After college, I had one dream: study at UC Berkeley, one of the best universities in the world. I never went, and this month I’m in Silicon Valley anyway.
In 2019, I joined a one-year program after graduation, mostly because Berkeley was one of its partners.
Berkeley is famous for studying the human brain, and I was sure this program was my way in.
My plan looked simple:
• Finish the one-year program first
• Get into Berkeley to study the brain
• Start my own company a few years later
Then one project in that program changed everything.
We were working with brain signals, and it introduced me to devices that gently send electrical signals to the brain to help treat conditions like depression.
The more doctors and patients I spoke to, the clearer it became that this treatment worked, but most people could not access it.
So I stopped planning to study the brain and started building for it.
That project became Marbles Health.
I never went to Berkeley, and today I’m a short drive away, pitching the company that detour created.
Brain health makes up 13% of what the world needs from healthcare.
It gets less than 5% of what the world spends on it.
That gap is exactly why this market keeps growing while most of healthcare has flattened.
• In 2000, the global mental health market was around $190 billion.
• Today it is $460 billion.
• By 2035, it is projected to cross $680 billion, roughly ₹60 lakh crore.
Here is what most people miss about that number.
Almost all of it goes to medication and therapy, the two tools we have relied on for 70 years.
They work, and they help millions. But for 6 out of 10 people, they only take the brain part of the way.
The next $200 billion of growth will not come from more of the same.
It will come from treatments that work with the parts of the brain we have only just started to understand, the electrical side and not just the chemical one.
That is the market we are building for.
Sources: WHO (share of disease burden), WHO Mental Health Atlas 2024 (mental health spending at about 2% of health budgets), IMARC Group (2025 market size), Towards Healthcare (2035 projection), STAR*D trial (treatment response). The 2000 figure is an estimate based on historical growth rates.
Clinicians read stroke on an MRI every day, but seeing the same lesion at cell level has been much harder.
On 11 October, PM Narendra Modi released an atlas that could help close that gap.
It's called C-STROKE, built at IIT Madras's Sudha Gopalakrishnan Brain Centre, which he also dedicated to the nation. The atlas comes from one stroke-affected brain:
→ Over 1,000 tissue sections carry detailed annotations.
→ The MRI scans link to micron-level tissue images.
→ Markers show inflammation, hypoxia and axonal injury.
→ The whole brain is reconstructed in 3D.
So you can pick a region on the scan and see what was happening in the tissue underneath.
The centre plans to map 100+ human brains the same way:
→ The maps will cover foetus to old age.
→ Some will come from brains with neurological disease.
At Marbles, we work with the living brain. Every neuromodulation decision we make depends on how well we understand the circuits we're reaching, so better maps help us and everyone else in this field.
None of this happens without the families who chose to donate, and I think they deserve as much credit as the scientists.
Nobody in the Valley has asked me, “Has anyone done this before?”
I’ve been here for a while now, pitching a second pillar of medical care alongside medication: portable, non-invasive neuromodulation with brain data monitoring.
The honest answer is that nothing like it exists yet.
Here, that’s where the conversation starts, not where it ends.
The response is “yes, and,” then “what would it take?”
The more ambitious the idea, the more people lean in.
Back home, one of the first things a founder is asked for is a precedent.
Someone who has already done it, so backing you feels safe.
The questions I hear back home:
• Has anyone done this before you?
• Who else has made this work?
The questions I hear in the Valley:
• Yes, and what else could it do?
• What would it take to build this?
I wish we saw more of the Valley’s energy in India.
I’m a little homesick for it already, and I haven’t even left.
She was 43, a homemaker in Hyderabad, doing everything her standard treatment recommended, and still getting worse every week.
Here is what her treatment looked like before EASE:
• Living with bipolar disorder for 1.5 years
• On medication since May 2025
• In a low phase since April 2026
• Five weeks of therapy before starting EASE
Despite all of this, her mood, sleep and appetite kept sliding.
Her psychiatrist, Dr. Priyanka Mahajan, added 20 sessions of EASE to her existing treatment.
Nothing else changed: 20 minutes, twice a day, for 20 days.
Her depression score dropped from 28 to 6, which is clinical remission (no longer clinically depressed), and her anxiety score went from 28 to 9.
By the halfway mark, the heaviest symptoms had already lifted. Her family noticed first.
She was sleeping and eating again, back at dinner, back in conversations, back to being herself.
Medication balances the brain’s chemistry, and EASE works on its electrical activity.
Together they reach what either one alone cannot.
Doctors often ask me, “Do you have a patient like mine?” Yes, here is one.
EASE is CDSCO-approved and used by psychiatrists across India.
If you are a clinician and want to see more cases like this, my inbox is open.
At 21, I rejected a job most people dream of.
At 23, I started a company during a pandemic.
At 25, I ran out of money and almost quit.
At 27, I figured it out just before it was too late.
At 28, Forbes put me on their list.
The timeline only makes sense looking backward.
As a neurotech founder, this is the Nobel Prize I have been waiting years to see.
This week, Karl Deisseroth, Peter Hegemann and Georg Nagel won the Nobel Prize in Medicine for optogenetics, and the story behind it is beautiful.
It started with a tiny green alga. Hegemann and Nagel found a protein inside it that opens up the moment light hits it.
Deisseroth then placed that protein inside nerve cells, and suddenly scientists could do something remarkable:
- They could switch brain cells on with light
- They could switch them off just as easily
- They could see which cells drive our behaviour
Using this tool, researchers even found the exact cells holding a mouse's fear memory, and brought that fear back with light alone, with no danger around.
But the part I love most is why it happened. Deisseroth is a psychiatrist who saw patients who were not getting better, and he wanted to understand what was really happening inside their brains.
That is the same gap we are working on at Marbles Health, taking brain science from the lab into the clinic.
Congratulations to all three of you.
"Our generation is getting dumber" is one of the laziest takes on the internet right now.
Every few weeks a reel goes viral quoting some IQ study and everyone panics.
So let me say this clearly as someone who studies the brain for a living.
IQ tests measure one very specific type of intelligence.
Pattern recognition, working memory, and processing speed.
That is it.
They do not measure creativity, emotional intelligence, adaptability, street smarts, or the ability to learn a new skill from a YouTube video in 30 minutes.
A farmer who cannot solve a matrix puzzle but can read weather, soil, and seasons is not dumb.
A 16-year-old who never memorized a phone number but built an app that 10,000 people use is not dumb either.
What actually happened since the 1990s is that intelligence shifted.
We stopped storing information and started learning how to access, filter, and apply it. That is not decline. That is adaptation.
Your grandfather memorized 50 phone numbers.
You know how to run a business from your phone.
Different skills. Different era.
Neither is smarter.
Stop measuring a fish by its ability to climb a tree and then calling the whole generation stupid.
I had stickers on my hostel cupboard at 19 that said "build a company the world uses."
Not a vision board or a Pinterest aesthetic.
Just handwritten stickers inside a cupboard in my single room at IIIT Delhi, arranged in a pathway like step one, step two, step three, leading up to that one line at the very top.
I never showed them to anyone.
Partly because I was not sure anyone would take it seriously, and partly because I was not sure I believed it myself yet.
But every morning I would open that cupboard to grab my things, and those stickers would look back at me before anything else did.
At that point I had no business background, no co-founder, no idea what neuroscience even was.
My father is an Air Force lawyer and my mother is a biology teacher.
Nobody in my family had ever built a company.
All I had was a feeling I could not explain and a cupboard that remembered it for me.
Six years later, the device we built at Marbles Health is :
→ Treating depression in 75+ hospitals across India.
→ Over 18,000 brain stimulation sessions delivered.
I still think about that cupboard sometimes.
The stickers worked.