When you publish something on a website like this one, the button doesn’t say “Publish.” It says Commit changes.
I’ve thought about those two words more than I’d like to admit.
The first time it really meant something was the SK hynix note. I had written, in public, under my own name, that I believe a stock can climb to $780 a share by 2032. I had also written that I own it. Everything was ready. The facts were sourced, the math was shown, the risks were laid out. And I sat there with the cursor over that green button, running through every way it could go wrong.
What if I’m early? What if I’m just wrong? What if someone who actually knows this industry reads it and laughs? What if, five years from now, the stock is lower and the note is still sitting there with my name at the bottom?
None of those questions had answers. They still don’t. I clicked it anyway.
There’s a line I’ve heard for years. I don’t know who first said it, and as far as I can tell, nobody does:
The edge of uncomfortable is where you find greatness.
I used to read that as motivation. Lately I read it as instructions. And the most important word in it isn’t “greatness.” It’s “edge.”
Most advice about discomfort makes the same mistake. It treats discomfort as one big thing you either avoid or charge straight into. Get out of your comfort zone. Jump. Burn the boats.
But the people who actually study learning describe something more precise. Educators have long used a simple map with three zones, popularized by the educator Tom Senninger (Mind Tools):
Psychology has an older version of the same shape. In 1908, two researchers, Robert Yerkes and John Dodson, found that pressure improved performance on hard tasks, but only up to a point. Past that point, it fell apart (Yerkes–Dodson law). Their subjects were mice, and scientists still argue about how neatly it applies to people. But the curve rings true for anyone who has ever frozen in an interview after nailing the practice run.
And when the psychologist Anders Ericsson spent his career studying how people become world-class at anything, he kept arriving at the same place. The best didn’t practice the most. They practiced right at the boundary of what they could do, a little past comfortable, never so far that they broke (Peak).
1 · The comfort zoneFlat ground. Everything here you already know how to do. It feels good, and nothing about it asks you to grow.
2 · The learning zoneThe ground starts to rise. It's harder, a little scary, but still within reach. You're stretched, and you can still think. The comfort, learning and panic zones, popularized by educator Tom Senninger.
3 · The edgeThe last step before it becomes too much. The view is best here, and so is the learning. This is where the quote lives.
4 · The panic zoneOne step past the edge, the ground falls away. Fear takes over, and you stop learning. You're just trying to survive.
5 · The curve hiding in the groundThe dashed line is the Yerkes–Dodson curve: on hard tasks, performance rises with pressure, peaks, then collapses. The landscape is that curve. Yerkes & Dodson, 1908
6 · The edge movesPractice at the edge long enough and the ground grows under you. Yesterday's edge becomes today's comfort, and a new edge appears farther out. Anders Ericsson's research on deliberate practice, Peak
So greatness isn’t on the far side of fear. It lives at a specific place: the edge of the learning zone, just before the panic. That’s the edge in the quote.
The problem is that from the inside, the edge and the cliff feel almost identical. Both make your stomach drop. Both make your mind invent reasons to wait. So most of us treat every edge like a cliff and step back to where it’s comfortable, and we call that being careful.
Comfort feels like the absence of a decision. It isn’t. Every year you stay where you already know the answers, the edge drifts a little farther away, and the thing you almost tried starts to feel like something for other people.
That used to be an acceptable trade, because the world moved slowly enough for a comfortable skill set to carry you a long way. When the tools change every few months, it doesn’t. Staying comfortable isn’t standing still. It’s drifting backward while everyone who’s stretching moves past you.
Here’s the trap I fall into most often. When something feels uncomfortable, I don’t avoid it outright. I research it. I plan it. I read one more article, watch one more video, outline one more version. It feels like progress. It looks like progress. It is not progress.
Brian Armstrong, the CEO of Coinbase, shared a lesson he says he learned from working with smart people, and it has stuck with me more than almost any advice I’ve heard (Brian Armstrong on X):
Action produces information.
His point: if you’re unsure what to do, do something, even if it turns out to be the wrong thing. The action itself tells you what you should actually be doing.
That idea quietly solves the problem from the last section. You can’t find the edge by thinking about it, because from a distance every edge looks like a cliff. The only way to learn where the ground actually drops off is to walk toward it and feel it under your feet. Thinking gives you opinions about the edge. Action gives you facts.
Overthinking is comfortable precisely because it produces no information. Nothing can go wrong in a plan.
The way I’ve started to think about it: every action is a question you ask the world, and the world always answers.
Publish the note, and you find out whether anyone reads it. Make the call, and the market eventually tells you whether you were right. Build the feature, and you find out it breaks on phones. None of those answers were available while I was planning. Every one showed up the moment I did something.
Most of what I’ve learned building Figmira came this way. The first research page I built had a flat map of stars. Looking at it is what made me want something more: a brain that grows with every note. That 3D brain only exists because the flat version was there to disappoint me. My first stock note opened with a dashboard of numbers, and reading it back showed me it buried the story; the rewrite started in a boardroom in 2002 instead. I wrote an early version of my last essay full of rocket ships and founder stories, read it again, and realized it sounded like everyone else. The rewrite became the most honest thing I’ve published.
Wrong actions aren’t wasted. They’re just answers you didn’t want. And an answer you didn’t want is still worth more than a question you never asked.
None of this means you should throw yourself at every scary thing. The edge is real, and so is the cliff. The skill is learning to tell them apart.
Here’s what I’ve noticed. At the edge, I’m uncomfortable but still curious. I’m nervous, but I can still think clearly enough to learn from what happens. At the cliff, curiosity disappears. All I can think about is getting out. I’m not learning anything; I’m just bracing.
Think of the hardest thing in front of you. How does it feel?
Pick the one that's closest. There's no wrong answer, only a place to stand.
So discomfort isn’t something to obey, and it isn’t something to ignore. It’s a signal to read. If you feel the stretch and you’re still learning, keep going. That’s the edge. If you feel nothing at all, you’re too comfortable; step forward. And if the fear has crowded out all the curiosity, take a step back. Not back to comfort. Just back to the edge.
Stepping back to the edge isn’t quitting. It’s how you stay out there long enough for it to matter.
Right now, my edge is this studio: learning in public how to write, build, research, and make calls that can be checked later. Some of those calls will age well. Some won’t. That’s the price of standing at the edge instead of watching from the comfort zone.
And I keep coming back to those two words on that button. Commit changes. Not “know how it ends.” Just commit to changing, and then find out what happens.
The edge of uncomfortable is where you find greatness. You won’t find it by thinking. Take a step, and let the world tell you where you’re standing.
Whatever you've been hovering over, here's some practice.
Now go find out what happens.
Iso, Founder of Figmira
]]>April 30, 2002. Hynix Semiconductor is barely alive. Memory prices have crashed, the banks it owes money to have taken control, and it’s carrying about $6.5 billion of debt.
On the table sits a lifeline from its American rival, Micron: Micron stock worth about $3.5 billion, plus a $200 million investment, in exchange for Hynix’s memory business. Taking it would make Micron the biggest memory maker in the world. It would also end Hynix.
The board votes. Unanimously, it says no. The CEO resigns (EDN). One headline at the time put it bluntly: “Hynix Hurts Itself by Rejecting Micron Offer” (TheStreet).
Twenty-four years later, that company, now called SK hynix, makes the memory that sits beside the most powerful AI chips on Earth. In just the last three months it reported, it earned $43.6 billion in operating profit. More than twelve times what Micron once offered for its memory business, earned in a single quarter.
That stubborn “no” is where our story starts. Since July 2026, anyone in the U.S. can own a piece of it through SKHY, its new Nasdaq shares. And we believe SKHY can run to $780 a share by 2032, about four times where it trades today.
Here’s our case.
Every AI model does two things: it thinks, and it remembers. The thinking happens in processors like Nvidia’s. The remembering happens in memory chips, and that’s where SK hynix lives.
There are two main kinds. DRAM is short-term memory: fast, and it forgets when the power goes off. NAND is long-term storage: slower, but it remembers. The most important kind today is HBM, or high-bandwidth memory: DRAM chips stacked like a tiny skyscraper and placed right beside an AI processor so data can move at enormous speed.
An AI chip without enough fast memory is like a genius who can only read one word at a time. The smarter the models get, the more memory they need. And AI is leaving the screen. Every robot that sees, moves, and learns will need memory of its own.
So, what if AI’s biggest bottleneck isn’t thinking, but remembering? We think it already is. Look at where the money is going:
Demand like this doesn’t show up in one company’s sales by accident. The world is building AI as fast as it can, and every piece of it needs somewhere to remember.
Only three companies make most of the world’s DRAM: Samsung, Micron, and SK hynix (Wikipedia). So why do we believe SK hynix is the one to own? Because it has been preparing for this moment for a long time, usually while everyone else was counting it out.
Hyundai founder Chung Ju-yung starts a chip business called Hyundai Electronics.
During the Asian financial crisis, Hyundai Electronics buys rival LG Semicon for $2.1 billion.
Memory prices crash. Banks that lent it money take control of the company.
Its board rejects a takeover offer worth about $3.5 billion, choosing to stay independent while carrying about $6.5 billion of debt.
Hynix is formally separated from the Hyundai Group.
South Korea's SK Group buys Hynix for $3 billion and renames it SK hynix.
SK hynix and AMD develop the world's first HBM, years before AI made stacked memory famous.
It starts mass-producing HBM3 for Nvidia systems, just before the AI boom takes off.
A memory downturn pushes it to an operating loss of $5.6 billion.
AI demand for HBM drives the most profitable year in its history.
Sources: Wikipedia (1983–2012); EDN, April 30, 2002 (the Micron offer); SK hynix (HBM history); SK hynix 2023 results. Dollar figures for 1998–2012 are as reported at the time.
Two decisions made this company. In 2002, it refused to be sold. In 2014, it bet on stacked memory years before AI made it essential. When the AI boom arrived, SK hynix was the one ready to supply it.
That head start now shows up in how much money it keeps from every sale:
Source: SK hynix Q2 2026 results; DRAM/NAND split from the earnings slides, via Investing.com. Dollar figures in this note convert Korean won at 1,388 won per $1 (Federal Reserve, September 18, 2026), one rate for every year so trends stay accurate.
Out of every $100 of chips it sold, $76 was left over as operating profit. For a business that runs giant factories, that is almost unheard of. It also holds far more cash ($63.4 billion) than debt ($13.4 billion), so it can keep building no matter what the market does.
And it’s building. About 10 major customers have signed long-term supply deals with SK hynix. Its newest AI memory, HBM4, started shipping this year. The M15X factory is speeding up, and a new cleanroom at its Yongin site opens in early 2027 (SK hynix). Companies don’t build new factories for demand they think is about to disappear.
Here’s the math. A share price comes from three beliefs multiplied together: how much a company sells, how much of that it keeps as profit, and how much investors will pay for each dollar of profit.
The biggest of the three beliefs is the last one. Today, investors pay only about 11 times earnings for SK hynix, because they’ve watched memory boom and bust for decades. Those long-term contracts are what could change their minds. A steadier business earns a higher price.
Delusional optimists still keep their eyes open. Believing in where SK hynix is going doesn’t mean ignoring the road.
Memory prices. This industry has always moved in waves. Here’s what the last six years looked like:
2021: $8.9B. 2022: $5.0B. 2023: a loss of $5.6B. 2024: $16.9B. 2025: $34.0B. 2026, first half only: $70.7B.
Source: SK hynix annual results (2021–22, 2023, 2024–25, 2026). Converted at 1,388 won per $1 for every year.
In 2023, SK hynix lost $5.6 billion. Three years later, it earned more than twelve times that in six months. The waves are real. They’re also why the upside is so large: when prices hold, the profits are enormous.
Samsung and Micron. Both are racing to sell more HBM, and Samsung is gaining fast:
Source: Counterpoint Research, via Seoul Economic Daily. Micron's first-quarter figure is implied from the reported change; shares are rounded.
SK hynix still sells half the world’s HBM. A market growing this fast can make room for all three, as long as demand keeps outrunning supply.
HBM4. Analysts are watching whether SK hynix’s newest AI memory earns as much as the chips before it (Seoul Economic Daily).
The world told Hynix its odds in 2002. It said no anyway. We think the story is just getting started.
Disclosure: I own SKHY shares. This note is my research and opinion, not financial advice. Do your own homework.
SK hynix Q2 2026 earningsThe latest quarter at a glance: sales, profit, cash, and what’s next→
Iso, Founder of Figmira
]]>Every few years, something that sounded unhinged becomes so ordinary nobody remembers it was ever up for debate.
Talking to a machine that talks back. Carrying the world’s libraries in your pocket. Earning a living from a bedroom with a laptop and a good idea. Each of those was, at some point, the kind of thing you’d say out loud and then quietly wish you hadn’t.
We have a word for belief that runs ahead of evidence. We call it delusion. And we use it right up until the evidence arrives, at which point we swap it for other words. Vision. Instinct. Obvious, really.
What the present calls it, what we call it once it works, and what everyone says in the end.
So when someone tells you you’re being delusional, listen to what they’re actually saying. They’re describing the present more than they’re describing you. They’re telling you that the world, as it stands today, has no room for what you’re picturing.
Of course it doesn’t. You’re picturing something that isn’t here yet.
“Be realistic” sounds like advice. A lot of the time it’s a forecast somebody else made about your life, handed to you as if it were a fact.
Realism comes from averages: what happened to most people, most of the time, in a world that looked like yesterday’s. There’s wisdom in that, and I’m not going to pretend otherwise. Averages kept a lot of people out of a lot of trouble. But an average describes a crowd. It has never described a single person in it.
The odds also carry a bias nobody mentions. They’re built mostly from people who stopped. Every “most people fail” includes everyone who quit in the first year, everyone who got talked out of it by someone who loved them, everyone who decided the dream was a phase. The statistic has no idea who you are. It only knows who gave up before you.
Use realism to measure the road. Just don’t let it pick the destination.
There used to be a deal. Most of us were raised on it. Be reasonable, work hard, don’t do anything crazy, and the future will take care of itself.
For a long time it mostly held. Of Americans born in 1940, about 90% grew up to earn more than their parents did at the same age. For people born in the 1980s, it’s about half (Opportunity Insights). The sensible path is still there. It’s just a coin flip now instead of a promise.
Then add this decade. The world is expected to pour more than a trillion dollars into AI this year alone (Goldman Sachs). Work is being redrawn in real time, and nobody, including the companies doing the redrawing, knows exactly where the lines will land. Ask people how they feel and you get a near-perfect split: about as many excited as nervous, and more and more, the same people feeling both.
When the ground is still, realism keeps you safe. When the ground is moving, realism can mean making very careful plans for a world that’s already on its way out. That’s the part that changed my mind. A little delusion used to be a personality trait. Right now it’s closer to a requirement.
People can feel the deal breaking, and they’re responding. Not by waiting.
Americans filed 5.62 million applications to start new businesses in 2025, far above the long-run average of about 3.5 million a year (Finder, citing U.S. Census data). At the same time, about a third of Gen Z now bets on sports or prediction markets, and roughly a third of Gen Z and millennials own crypto. Among people putting money into high-risk bets like these, 73% say they’re doing it because they feel financially behind, and among Gen Z it’s 80% (Northwestern Mutual, 2026). Some writers have started calling it “financial nihilism.”
I understand the impulse completely. When the slow road stops looking like it goes anywhere, a fast one starts to look reasonable.
But I want to be honest about what separates those bets, because this is where the whole idea lives or dies. A parlay is a bet where the odds are fixed and the house wrote them. Nothing you do after you place it changes anything. A bet on yourself is different. The odds move. They move every time you learn something, ship something, fix something, try again. Starting a business, learning a skill from scratch, putting your work in public: those are bets where effort gets a say.
So yes, I think more people should be taking real chances right now. Just the kind where you’re holding the dice.
What does that look like on an ordinary Tuesday? Usually it’s not dramatic. It’s an hour after work spent learning the thing everyone says you’re too late for. It’s publishing something small before it’s ready, with your name on it. It’s saying out loud what you believe will happen, so you can be proven wrong. It’s keeping the job that pays the bills while you build the thing that might replace it. Delusion, most days, looks a lot like homework.
The hardest part of believing in something early is the gap.
There’s the version of your life you can see in your head, and the version everyone else can see right now, and for a long time the two don’t match. You live in the space between them. You make decisions based on a future no one else can confirm. You explain yourself at dinner and watch people nod politely. Some days, you’re the one nodding politely at yourself.
No proof yet.
Polite nods at dinner.
"Is this just a phase?"
Some days, you nod too.
Keep the picture lit.
Then one day, the two lines meet.
That gap is where most dreams die, and usually it isn’t a dramatic failure that kills them. It’s fatigue. Believing without proof, day after day, until it’s easier to let the picture go and rejoin the present.
The people who make it through still doubt themselves. They just stop letting the doubt make the final call. Certainty only ever shows up at the end, after the thing works. If you wait for it before you move, you won’t move.
There’s a kind of delusion that wrecks people, and this essay isn’t an argument for it.
The dangerous kind lies about the present. It ignores the bank balance, the bad reviews, the broken code, the thing that plainly isn’t working. It confuses wanting something with having it.
The useful kind looks straight at every hard fact about today and still refuses to let those facts decide tomorrow.
The best illustration I know comes from James Stockdale, a U.S. Navy pilot who was shot down over North Vietnam in 1965 and held as a prisoner of war for more than seven years, much of it in brutal conditions. Decades later, the writer Jim Collins asked him who didn’t make it out. Stockdale’s answer surprised him: the optimists. The men who said they’d be home by Christmas, then by Easter, then by the next Christmas, and slowly lost heart as each date passed.
Stockdale survived by holding two things at once. Absolute faith that he would get out and that the experience would mean something. And complete honesty about how bad things were right now. Collins named that combination the Stockdale Paradox.
Impossible about the dream. Ruthless about the facts.
That’s the line I’m trying to walk. Out of my mind about where this goes. Completely sane about where it is today.
I never believed I could build something like this.
I’m not an engineer. I’m not a designer. I’ve never worked in a newsroom or a research firm. For most of my life, the kind of website you’re reading right now belonged in the category of things other people make: people with teams, with training, with years of practice I didn’t have.
Then AI changed what I thought was possible for someone like me.
It started small. I’d ask questions I would have been embarrassed to ask a person. What’s a stylesheet? Why did that break? How do you make a number count up when someone scrolls to it? The answers kept coming, patiently, at any hour, and slowly the list of things I couldn’t do got shorter. A doorway on the home page that opens like a gateway into another world. Charts that come alive as you read. A research page where every note becomes a neuron in a brain made of stars. A word on this very page that rewrites itself from “delusional” to “obvious.”
Newsrooms build pieces like this with teams of designers and engineers. I’m building mine with curiosity and an AI that never gets tired of my questions. The result, when it works, feels less like a website and more like a picture book assembling itself while you read.
It doesn’t always work. I’ve wiped out my own home page with one wrong upload. I tried to create a mascot for this studio and had to shelve him because I couldn’t draw him the way I saw him. I rewrote my first stock note from the ground up more than once before it read like a story instead of a report. Every one of those misses is still in the history of this site, and I’m fine with that.
What I’m risking isn’t dramatic, but it’s real. It’s my evenings. It’s my name on public calls that could look foolish later: I wrote that I believe SKHY can reach $780 a share by 2032, and I own the stock. If I’m wrong, the note stays up.
Here’s why I do it anyway. If AI could make someone like me believe I could build this, it can do the same for a lot of people who were told, or told themselves, that building wasn’t for them. The least I can do is share what I’m seeing. That’s what Figmira is: my delusion, out loud, pointed at the things I find most fascinating right now. AI, which made all of this possible for me. Robotics, where that intelligence is starting to move through the physical world. Crypto, where people are rebuilding how money and ownership work. And investing, where you find out whether you believed in the right things.
Every note labels its facts as facts and its beliefs as beliefs. I think that’s what delusion looks like once it grows up.
The world will tell you your odds. Be out of your mind anyway.
Iso, Founder of Figmira
]]>Building a world takes more than imagination, though. It takes understanding your materials. Right now, four of those materials are changing faster than anything I’ve seen: artificial intelligence, blockchain, robotics, and the way capital moves toward new ideas.
So before Figmira Studios builds anything, it’s going to study. These notes are where that work lives.
Artificial intelligence. How intelligent systems learn, create, and work alongside people. I’m less interested in whether AI replaces us and more interested in what we can build with it that neither of us could build alone.
Blockchain. Open networks and digital ownership, without the noise. What works today, what doesn’t work yet, and what’s worth building on.
Robotics. AI leaving the screen. Machines that sense, move, and act in the physical world, and what it takes to build them.
Investing. How capital finds the ideas that shape the future. I’ll share how I think about evaluating new technologies and the companies behind them.
One note on the investing pieces: they’re my research and my opinions, not financial advice. Always do your own homework.
The worlds Figmira wants to build someday will be made from these technologies. Understanding them deeply now is how we build something that lasts later.
Every realm on this site is still a figment. These notes are the first step toward making them real.
Iso, Founder of Figmira
]]>