Chapter 176: Human Neuron

“As you know, all the computational modules required for AGE are currently being produced at the Dongtan Plant 2.”

I nodded at Director Cha Dong-hyuk’s words.

Dongtan Plant 2 was a facility designed with semiconductor production in mind from the very beginning.

At this point, it was the only place capable of producing semiconductors.

Executive Vice President Hong Chung-ki added further explanation.

“However, if production volume increases beyond the current level, Dongtan Plant 2 alone will struggle to cope.”

“You mean there isn’t enough space?”

“Yes, when the factory was originally designed, half was allocated for smartphone production lines, and the other half for semiconductor lines……”

“Ah, I remember. Some sections were shifted to smartphone production because semiconductor production was delayed.”

Smartphone production is still capacity-constrained.

“So, while Dongtan Plant 2 can handle small-scale production, it becomes challenging to expand if volumes increase.”

Knowing this, I had already instructed the construction of a new plant in Pyeongtaek.

I turned my attention to Park Cheol-jin.

“How’s the progress in Pyeongtaek?”

Before reporting, Park Cheol-jin displayed the information on the screen.

“Currently, Pyeongtaek is operating two separate sites.”

The materials Park Cheol-jin presented clearly outlined the locations and progress.

[Site 1] Data Center Construction

Location: Pyeongtaek West Industrial Complex
Scale: 32,000 pyeong, 5 floors above ground/2 below ground, total floor area of approximately 180,000 square meters
Construction Period: Phase 1 to be completed in 4 months

[Site 2] Semiconductor and Smartphone Production Line Construction

Location: Pyeongtaek East Advanced Industrial Complex
Scale: 96,000 pyeong of land, including 5 cleanroom lines
Construction Period: 80% of structure completed, equipment installation scheduled in 3 months.

“All the buildings are made with ultra-high-strength concrete, right?”

Park Cheol-jin nodded.

“Yes, you instructed us to do so during the planning phase. Only ultra-high-strength concrete is being used for the data center and all manufacturing line buildings.”

“What about the cost?”

“I’ll address that.”

It was Executive Director Oh Seong-hak.

“The budget allocated so far for the construction of Pyeongtaek Plants 1 and 2 is 430 billion and 580 billion won, respectively. If there are no design changes, we expect to complete the project without additional funding.”

“Was there no cost increase due to using ultra-high-strength concrete?”

“There was…… but not significant.”

Executive Director Oh Seong-hak displayed separate data showing the extent of construction cost increases compared to regular concrete.

“Certainly, producing ultra-high-strength concrete in-house has kept the burden manageable.”

If we had sourced it externally, costs would have surged by over 50%.

Thanks to in-house production, we’ve kept it within a 5% increase.

“At this rate, we can standardize all future buildings with ultra-high-strength concrete.”

“We’ll review that.”

Honestly, even 5% isn’t a small figure.

But since we’re building, I wanted to make them as sturdy and proper as possible.

So, while I couldn’t be on-site full-time like during the NCC project,

I made it a point to visit whenever I had time, even if just to do some simple digging.

‘No need to conserve Midas’s abilities.’

I turned back to Executive Vice President Hong Chung-ki.

“Regarding semiconductor equipment, aside from the devices Team Leader Park Sang-bae is developing, what’s the status of ordering commercial equipment?”

It’s natural that equipment under development takes time.

But I had already instructed the procurement of commercially available equipment in advance.

Executive Vice President Hong Chung-ki showed me the list of related equipment.

The data was organized into front-end and back-end processes.

[Front-End Equipment]

Photolithography Equipment

· Purpose: Pattern Formation

· Supplier: ASMU

· Quantity: 5 Units

· Cost: 790 Billion Won

Etcher

· Purpose: Circuit Pattern Etching

· Supplier: TELL, RAM Research

· Quantity: 8 Units

· Cost: 230 Billion Won

CVD/PVD System

· Purpose: Thin-Film Deposition

· Supplier:…….

-…….

[Back-End Equipment]

Die Bonder / Wire Bonder

· Purpose: Chip Bonding and Connection

· Supplier: ASMP/Kulicke&Sofa

· Quantity: 12 Units

· Cost: 53 Billion Won

Encapsulation Line

· Purpose: Chip Packaging and Sealing

· Supplier:…….

Test Handler

· Purpose: Function/Yield Testing

· Supplier:…….

The list was dizzyingly extensive.

The document, including even small test boards, exceeded a hundred pages.

Just skimming through it took considerable time.

“……So far, about 3.2 trillion won has been invested in equipment procurement alone.”

Executive Vice President Hong Chung-ki nodded.

“For reference, this list is based on a capacity of 2.8 million units.”

“Hmm…… We still have a long way to go.”

It can’t be perfect from the start.

In fact, 2.8 million units is ridiculously insufficient compared to what’s needed.

Just to fully replace the Quantum series with my designed semiconductors, we’d need at least 20 million units per month.

The current situation covers barely 10%.

‘To produce high-performance AI GPUs, how much more investment will be needed…… I can’t even fathom.’

Semiconductors being a ‘money pit’ isn’t an exaggeration.

“Are funds for equipment and semiconductor development sufficiently secured?”

Executive Director Oh Seong-hak’s pupils flickered momentarily.

“……There’s some leeway right now, but I wouldn’t say it’s ample.”

“Not ample?”

It was hard to understand immediately.

Our main cash cow, the mobile business, was fully on track after Ahsung Electronics withdrew.

The profit from that alone exceeded 5 trillion won monthly.

Adding Dojin Soft’s Lunar, KTalk, and various monetization models, the monthly operating profit was over 3 trillion won.

“Yet, you’re saying funds are insufficient?”

Executive Director Oh Seong-hak sighed softly.

“……As I said, it’s not insufficient, just not ample. However, if investment areas increase further, the situation could change.”

“That many areas are draining funds?”

Executive Director Oh Seong-hak shared his laptop screen on the monitor.

[Dojin Group Funding Allocation]

Semiconductor Equipment Development

High-Speed TSV Aligner

Function: Real-time TSV (Through-Silicon Via) alignment and scanning for 3D integrated circuits
Features: Alignment accuracy within ±10nm, equipped with AI-based real-time correction
Development Cost: 2.82 Trillion Won
Note: Addresses high-density TSV arrays for AI SoCs, impossible with existing equipment

Photonics Assembly System

Function: Automated assembly of silicon photonics-based optical computing units (optical interfaces)
Features: Equipped with ‘non-uniform lens-wavelength matching system’ to minimize optical signal interference
Development Cost: 3.5 Trillion Won
Note: Essential for high-bandwidth Photonic NPUs

Memristor Computing Array

Function: Analog computation for…….
Features: Irreversible at the cell level…….
Development Cost: 3.93 Trillion Won
Note: To be used as a core unit in LLM inference accelerators

ReRAM Cell Staggering System

Function:…….
Features:…….

LUNA-EVX (Exposure Equipment for Ultra-Fine Circuit Printing)

Function: Printing circuit patterns with 0.3nm-level extreme resolution
Features: 4x resolution and 7x precision compared to ASMU’s EUV equipment under development
Core Technology: Photon Focusing Technology, Dual-Sided Active Interface Masking System
Development Cost: $23 Billion (approx. 25 Trillion Won)
Note: ASMU’s EUV can implement up to 7nm half nodes; LUNA-EVX…….

The moment the figures and details were projected on the screen,

Gasps filled the conference room.

“……How much is all that?”

“We’re pouring 40 trillion won into semiconductor equipment development alone?”

I knew to some extent.

But seeing it organized like this, the sheer scale of investment hit home.

“What is the LUNA-EVX equipment that requires so much funding?”

A staggering 25 trillion won for a single piece of equipment.

Park Cheol-jin’s curiosity was understandable.

“LUNA-EVX is used in the most critical semiconductor process—drawing ultra-fine circuits.”

“Ultra-fine circuits? Just how fine can it draw to warrant such massive development costs? I hear there are already similar equipment available……”

Why not just buy and use them?

But that question came from not understanding the level of semiconductors I aimed to create.

Chilbong, sitting beside me, continued the explanation.

“The most crucial aspect of electronic circuits is ‘density.’ It’s about how many components can be precisely arranged in the smallest space.”

Chilbong manipulated the computer, displaying two circuit images on the screen.

The left was a 45nm-level circuit.

The right, a 7nm-level.

Clearly, the right one was far more intricate and complex.

“Drawing smaller circuits isn’t just about making ‘smaller components.’ It means processing calculations faster, handling more, with less power.”

“That’s why everyone emphasizes the importance of advanced processes. And LUNA-EVX, the exposure equipment, enables that.”

“Correct. Exposure equipment is like the ‘brush’ that draws circuits on silicon wafers. The highest-spec equipment currently recognized is ASMU’s 7nm-level EUV.”

As Chilbong finished, the conference room buzzed with awe once more.

Those who hadn’t fully grasped the data now understood just how remarkable LUNA-EVX was.

“So…… the equipment we’re developing can achieve a resolution of 0.3nm?”

The air in the room shifted subtly, and no one spoke easily.

“0.3 nanometers—that’s almost atomic-level, isn’t it?”

Executive Vice President Hong Chung-ki’s words made Chilbong nod.

“Yes, it was thought impossible due to the physical limit of ‘light wavelength.’ But combining the photon focusing technology and dual-sided active interface masking you designed, we expect to surpass that limit.”

I added to the explanation.

“Existing equipment exposes only one side at a time, but LUNA-EVX draws circuits on both sides simultaneously.”

This dramatically increases precision, boosting integration density by over fourfold and power efficiency by up to sixfold.

“Of course, it won’t achieve that performance right away.”

There will be trial and error, and much to adjust initially.

But one thing was certain.

If this equipment is successfully developed, I won’t just create ordinary AI semiconductors.

I could implement sub-3nm ultra-high-density cells and even 0.5nm-level artificial synapses.

In theory, this means circuits mimicking human brain neurons exactly.

In other words, we could reach the level of drawing ‘human neurons’ on silicon.