Chapter 249: Milo (6)
I stepped out onto the terrace and immediately called Alex.
What he told me was simple.
[Stark suddenly showed up and asked if we needed investment money. He even pulled out a $100 million check.]
Without any prior notice, Stark barged in, showed the check, and even made an outrageous offer, saying he would invest over a billion dollars if it was for "ethical AI development."
But, [Of course, I turned him down. Because I had a promise with Sean.]
That’s right.
I had already provided Alex with ample funding, under the special condition that he would not collaborate with Stark in any form.
Alex kept that promise.
However, the problem came next.
[After I refused, Stark just smirked and said, “Ah, it looks like the Saint has already blessed you.”]
"The Saint."
That’s my nickname in WSB.
In other words, Stark had already figured out that I was the investor.
“So, what did you say in response?”
[At that moment, I just acted like I didn’t understand what he meant. Then he laughed again and asked, “Did you team up with someone who can perform miracles or predict the future?”]
Instead of directly asking "who the investor is," Stark kept throwing memes related to me, making things awkward for Alex.
...
[・・・It’s no exaggeration, he rattled off over thirty memes in a row. From asking if I was a marine mammal living in a black-and-white-spotted sea to if I enjoyed shark fin soup... There were so many that were hard to understand. He really is a unique guy, just like they say.]
At this point, it was safe to assume that he was already convinced the investor was Orca.
[I tried to play it off somehow, but he seemed pretty certain. Honestly, it was inevitable. These days, Sean’s name is just so hot in Silicon Valley.]
“Me?”
[Yes, after the Moonshot Project, there isn’t anyone in the AI industry who doesn’t know Sean.]
‘Ah, right, that.’
The "Moonshot Project" mentioned here refers to the plan I announced the last time I visited Silicon Valley.
Back then, I declared I would invest $1 billion into a startup integrating AI technology for treating Castleman disease.
But,
[Originally, autonomous driving and chatbots were the big trends, but after that, a lot of companies started shifting to healthcare.]
In 2015, the AI industry wasn’t that big yet.
And I had dropped a massive $1 billion into that small pond.
In other words, I naturally became known as the biggest heavyweight investor in the AI field.
[Under those circumstances, if I turn down a $100 million check and even reject another $1 billion investment offer... of course they’re going to think of Sean.]
Just as Alex pointed out.
‘So much for trying to avoid it.’
I had hoped to avoid getting entangled with Stark as much as possible, but maybe that was wishful thinking.
Since we’re interested in the same technology, it was only natural that our paths would cross.
‘Well, that doesn’t necessarily mean we have to clash.’
There were still ways to avoid him.
But that was something for later.
Right now, I had more urgent and important matters piling up.
In the end, I stayed in Philadelphia for four more days.
Partly to attend Milo’s funeral...
But also because there was something important I had to take care of first.
One of those things was securing a sample of Milo’s biological tissue.
However, that wasn’t easy.
“To get that... we’d have to perform an autopsy, right?”
Milo’s parents firmly refused.
After all, it meant cutting into their child’s body.
“Please, just let our child rest now. He was just a little kid...”
Their reaction was entirely understandable.
But we also had a reason we couldn’t back down.
“It’s precisely because Milo was a child that we can gain crucial clues.”
Data from a child gives us far more hints than data from an adult.
In fact, Castleman disease is extremely rare in children.
Most patients only develop it once they’re adults.
Which suggests that the disease is likely triggered by acquired factors.
Things like infections, environmental factors, or unhealthy lifestyle habits slowly damage different parts of the immune system, eventually creating Castleman’s unique “madness” switch.
“With current sequencing technology, we can trace the damaged gene regions to some extent. Based on that, we might even uncover the disease’s fundamental cause.”
“But can’t you get that kind of sample from other patients?”
“Unfortunately, no. In adults, the damage is so widespread that it’s hard to trace. But in Milo’s case...”
As humans age, they’re exposed to more environmental factors, and various areas inevitably get damaged.
But Milo?
Being so young, he had relatively little external damage.
In other words, if an adult’s sample has over a hundred areas of damage, Milo’s body might have at most twenty.
This means we could dramatically narrow down the range of the “madness” switch.
Rachel carefully explained all this step by step, but even then, the parents still looked hesitant.
“There are already tissue and blood test records, right? Isn’t that enough?”
“Sadly, it’s not. For this analysis, we need specific tissues like lymph nodes, the spleen, and bone marrow.”
“But Milo... it just feels so pitiful... I don’t think we can do it.”
At that moment, Rachel’s eyes changed.
“I don’t want to remember Milo as just a poor, pitiful child.”
Her voice was trembling, but her tone was firm.
“Milo wanted to become strong more than anyone else. Like a Tyrannosaurus.”
She seemed choked up and took a moment to steady herself before continuing.
“To us adults, it might sound like a silly dream, but for him, it was deadly serious. And right now, Milo has the power to save more people than any adult. I truly believe... Milo would have wanted this. He wanted to become strong, and he was a child who showed courage to the very end.”
It was a pretty self-serving interpretation...
But in that moment, I remembered Milo vowing, “I’ll become a T. rex.”
If Milo had always been terrified and crying, this kind of persuasion would have been impossible.
But back then, he clearly overcame his fear and showed his will to become stronger.
His parents must have remembered that too, because they looked at each other and then burst into tears.
After a long while.
Finally, they nodded.
“Yes, we agree.”
I got the answer I wanted, but I didn’t feel at ease.
So, in my own way, I did everything I could for Milo’s family.
First, I stayed at the funeral from beginning to end.
Many people recognized me and came up to talk, and I responded kindly to as many as I could.
That "Ha Si-heon" was postponing all his work to stay by the child’s side...
That alone must have meant something to some people.
Not only that, but I also arranged a new home for Milo’s parents.
It wouldn’t be easy for parents who lost a child to keep living in a house full of memories of him.
And I decided to name the ward I plan to donate to the University of Pennsylvania Hospital “Milo T-Rex.”
Lastly, I went back to the tattoo shop I had visited before and had Milo’s name inked on the inside of my wrist.
This time, though, I asked for something extra.
“Please add a Tyrannosaurus too.”
It was a tiny T. rex, about the size of a fingernail, but it looked quite proud and strong.
Almost as if it were shouting, “I don’t need your pity.”
Of course, there’s no way Milo actually had such deep thoughts.
A three-year-old wouldn’t even understand a complicated concept like “pity.”
In the end, it was nothing more than a rationalization to ease my own guilt.
But,
—I’ll become a T. rex!
Milo’s shout back then allowed me to justify it in this way.
This little dinosaur on my wrist was my own way of expressing gratitude for that permission.
As soon as I returned to New York, I threw myself into work without a moment to breathe.
The most urgent task was analyzing Milo’s sample.
Through it, I had to find Castleman’s “madness” switch.
However...
“The current capture efficiency of RNA sequencing technology is only around 5–10%.”
The problem was the technical limitations.
Even if we started analyzing right now, it would be difficult to obtain even 10% of the total gene expression information.
Ultimately, it meant that proper analysis wouldn’t be possible until the technology advanced further.
“Besides, I heard that the spatial transcriptomics technology you mentioned is currently only being researched in Sweden...”
RNA sequencing analyzes all RNA within cells, so you can’t tell where each cell is located in the body.
But spatial transcriptomics captures that spatial context as well.
This technology was absolutely necessary to find the “madness” switch...
‘So I’ll have to wait another year or two?’
This technology wouldn’t really start advancing until after 2017.
Right now, there was only one small startup in Sweden trying to commercialize it, but they were moving slowly due to a lack of funding.
There was only one way to speed things up.
“Let’s acquire it through Quantum Genome.”
Quantum Genome was one of the companies I invested in, and it would soon grow into a leading name in spatial transcriptomics.
If we could integrate the Swedish technology there, the research would undoubtedly accelerate.
“They’ll probably welcome it too. They’re probably struggling with acquisition costs, so let’s offer to fund them. Of course, we’ll need to get shares in return.”
This would solve the gene analysis issue to some extent.
But the real obstacle was elsewhere.
“Even after the analysis, we’ll need AI to trace the immune pathways and pinpoint the exact information.”
The “madness” switch doesn’t occur due to a single gene abnormality.
It’s a phenomenon caused by a complex combination of multiple mutations.
To uncover this, we’d have to analyze billions of data patterns to find hidden correlations.
And for that, deep learning technology was essential.
However, in 2015, deep learning was not yet mainstream in the AI market.
Only some big tech companies and academia were conducting small-scale research.
‘I guess I’ll have to advance this by a year or two too.’
This wasn’t a problem that could be solved with money alone.
First, the hardware needed to support it.
“With the chips currently on the market, it’s impossible to handle the scale of deep learning computation Sean mentioned.”
In AI, the most important factor is computational speed, which depends on GPUs.
The company that practically monopolized the GPU market was Envid.
But their main products were for gaming and lacked functions optimized for deep learning.
“The Maxwing architecture has limitations. In a few months, they say a Parsa architecture product will be released...”
What I needed was the next generation, the Bolton architecture.
That lineup wouldn’t be released until 2017.
Only then would Tensium cores be introduced, enabling the FP16 computation acceleration I needed...
And that was something only I knew.
At this point in time, the term “FP16 computation acceleration” didn’t even exist yet.
“We’ve explored the possibility of optimizing GPUs for deep learning, but we haven’t received a clear answer. From Envid’s perspective, it would be risky to pour resources into a product without guaranteed demand.”
The deep learning market was still too small.
Meanwhile, gaming products accounted for over 80% of Envid’s revenue and were their main cash cow.
Naturally, they prioritized gaming GPUs.
However, from my standpoint, I needed to make them focus more on producing GPUs for deep learning...
This was no easy task.
‘They’re a publicly listed company.’
Most of the companies with gene analysis technology were private.
In private companies, as a major shareholder, I could simply persuade the executives, and things moved quickly.
But Envid was a public company.
And I only held an 8% stake.
I was a major shareholder in name, but that wasn’t enough to move the executives.
“We’ve scheduled a meeting with the CEO for now, but it’s uncertain how much they’ll cooperate.”
The chance that the CEO would unconditionally accept my proposal was low.
If he recklessly changed business directions based on my words alone, he would face backlash from other shareholders.
In the worst case, he might even face pressure to step down.
“At this point, there’s no reason for them to prioritize GPUs for deep learning...”
Not wrong.
But.
“It’s fine. Let me ‘persuade’ them.”
If a reason doesn’t exist, I’ll just create one.
Having made up my mind, I immediately flew to California to “persuade” them.
I arrived at 11 p.m.
But just as I was walking into the hotel lobby I had reserved...
“Sean?”
Someone called my name.
I turned my head to see a man rising from a lounge sofa and walking toward me.
I had never met him in person, but he was so famous in my past life that I recognized him instantly.