Chapter 235: Lunar Probe

“I’ll give the report on the mobility sector.”

Park Cheol-jin flipped to the next slide, and neatly organized text and numbers appeared before our eyes.

[Mobility - Quarterly Performance Summary (Provisional)]

(1) Prime (Premium Electric Vehicle)

• Units Sold: 1,000 (All first reservation units delivered)

• ASP (Average Selling Price per Unit): $2 million

• Revenue: $2 billion

• Estimated Operating Profit: $680 million (Operating Margin 34%, Initial Mass Production Incentives, Parts Deductions Reflected)

“Delivery of Prime (PRIME) is complete, with 700 units delivered in bulk and 300 units delivered sequentially. Refurbished and parts kit revenue will be gradually reflected from next quarter onward.”

I nodded at Park Cheol-jin’s report.

As a premium vehicle, both the unit price and margin were satisfactory.

“What about the second reservations?”

“We’re planning to open them soon.”

Peak output of 1,550 horsepower, 0-60 in 1.69 seconds, top speed of 457 km/h, and a range of 680 km.

Purchase inquiries for Prime, which boasted overwhelming specs and a sleek design, were nonstop.

Naturally, I expected the second reservations to sell out instantly as well.

But.

“To secure autonomous driving technology, launching the EV-Nova, a mass-market vehicle, is more urgent than Prime.”

Real-time driving data was absolutely necessary to complete autonomous driving technology.

Not data from just one or two vehicles.

We needed real-time data from tens, even millions of vehicles, encompassing all weather conditions, times of day, and road environments.

“EV-Nova is scheduled for normal shipment from next quarter.”

Perhaps knowing I had a keen interest, the progress on EV-Nova was recorded in relatively detail.

(2) EV-Nova (Mass Production Plan)

• Shipment Period: Q2 2018

• Current Production Line CAPA: 3,000 units/day (based on 3 shifts)

• Process Yield (Initial): 92.4% → Target 97%

“Preparations for mass production are complete, and we entered full-scale production last week.”

“First reservations were for 580,000 units, right?”

“Yes, to be exact, 582,798 units.”

Thanks to acquiring Ahsung Motors’ electric vehicle division, we were able to bring forward the EV-Nova delivery schedule by over a year.

Even so, it would take at least six months to supply all units.

“I hear purchase inquiries are pouring in even now. What’s the plan for additional reservations?”

At Oh Seong-hak’s question, Park Cheol-jin’s pupils flickered momentarily.

But soon he calmly continued his explanation.

“Once we start delivering EV-Nova, we plan to open the second reservations immediately.”

“If we take reservations in this state, won’t the wait times become too long…?”

With the current CAPA, it would take half a year just to fulfill the secured orders.

Adding more reservations would mean wait times ranging from six months to over a year to receive a vehicle.

“To bring forward delivery times, we need to expand CAPA. However……”

“If it’s about costs, don’t worry and tell me.”

Park Cheol-jin, who had been hesitating due to CAPA expansion costs, brought up the additional slides as if he’d been waiting.

[CAPA Expansion Plan]

• Target CAPA: 3,000 units/day → 5,000 units/day

• Expected Investment Cost: 1.5 trillion won

(Terra Casting, Ultra-High Strength Frame Line, Heat Treatment, Automation Equipment, etc.)

• Investment Execution: Construction starting in Q3, completion within 6 quarters (Target)

“Terra casting is definitely faster, but the initial investment cost is quite high.”

“Still, once setup is complete, the investment can be recovered within a year.”

Knowing this already, I nodded in agreement.

“First, I’ll support the costs needed for CAPA expansion by purchasing Dojin Frontier bonds. Proceed immediately.”

Right now, it might seem like a money pit.

But with Prime (PRIME) sales already started and EV-Nova shipments ramping up, a stable structure would be established without needing additional funding.

Moreover, additional investment was something we had anticipated to some extent.

The problem was that this wasn’t the only place needing more money.

“How’s the preparation for autonomous driving going?”

This time, instead of Park Cheol-jin, Director Kim Se-bin took the microphone.

“As you well know, sir, the core of the autonomous driving technology we’re developing consists of two things.”

“You’re referring to Camera Only and the cloud learning structure, right?”

“Yes, that’s correct.”

On the meeting room screen appeared a diagram of 12 cameras surrounding the vehicle’s exterior in a 360-degree ring and a pipeline transmitting data to the cloud.

[Autonomy Stack - ‘Aquila v0.9’]

• Sensor Configuration: Multi-Ultra Wide Angle + Telephoto Camera (Visible Light + NIR), IMU, Wheel Odometry, GPS. (No LiDAR / Radar)

• Perception Prediction Planning: End-to-End Neural Network + RSS Compliant Stability Constraint Module

• Learning System: Cloud Feedback Based

• Communication Infrastructure: Currently 4G/Wired Backhaul → Future Transition to Luna Link Backhaul

• Launch Target: March 2020

“Are there no issues with autonomous driving using only cameras?”

Hesitating momentarily at my question, Kim Se-bin answered with a serious expression.

“Actually…… there are quite a few limitations to implementing autonomous driving without LiDAR and radar. The difficulty of camera-based technology increases sharply in situations like backlighting at night, heavy rain and snow, headlight flare, brightness differences at tunnel entrances and exits, lost lane markings, and most importantly, absolute distance estimation and occlusion handling.”

I was already aware of this.

But I drew a firm line.

“The most important thing in autonomous driving is scalability! LiDAR and radar not only require expensive hardware costing over $100,000 per vehicle but also come with additional constraints in procurement, mounting location, and power consumption. There are too many scalability issues.”

Especially for cloud-based learning, the most important thing is the volume of data.

Who would buy a vehicle with LiDAR and radar costing over $100,000?

The performance difference between autonomous driving technology trained on data from a thousand vehicles versus millions is incomparable.

Chilbong, who had been listening silently until now, joined the conversation.

“So we’re planning to redesign the stack composition as image → 3D occupancy space → vector space planning. For absolute distance measurement, we’ll combine monocular depth and multi-view parallax estimation to compensate.”

“What about issues like night, backlighting, and bad weather?”

“For those, we plan to raise the SNR using a combination of NIR (near-infrared) channels and polarizing filters.”

“Not bad. Tunnel and headlight flare issues can be easily resolved by securing learning data coverage.”

Chilbong nodded, seeming to have thought the same.

Director Kim Se-bin then offered an additional opinion.

“Then how about equipping only cameras on mass-production vehicles and attaching LiDAR and millimeter-wave radar as ‘shadow sensors’ on some development vehicles to use for label verification and safety cage operation?”

“So use only camera signals on mass-production vehicles and utilize LiDAR and radar data from development vehicles as reference answers?”

“Yes.”

It was a good idea.

When I indicated approval, discussion immediately followed regarding camera shape, placement, and procurement methods.

Cameras were the first business I started after founding Dojin Tech, and it was a field where we possessed world-class technology, so the technical discussion concluded quickly.

“Are there any cost-related issues?”

“Autonomous driving technology development is in collaboration with Han Yu-ju’s team at Dojin Soft, and most related costs are covered by them.”

Dojin Soft was one of the core pillars of this project, jointly developing autonomous driving software and Luna.

Moreover, with Lunar, KTalk, Feed, and various other monetization platforms already operating stably, they had ample financial capacity.

“What does Director Han say?”

Director Kim Se-bin answered my question.

“This Aquila stack is fully compatible with the autonomous driving module of the Luna platform, and Director Han said they’ll integrate the OTA structure and vehicle communication interface by year-end. That way, a beta rollout should be possible by next quarter.”

Chilbong added from the side.

“Two Luna Link satellite gateways are already installed at the Goheung base, and the remaining three are planned for Seosan, Pohang, and Ulsan. With EV-Nova development vehicles uploading driving data primarily over 4G and offloading communication traffic to Luna Link during nighttime, there shouldn’t be major issues with data acquisition.”

I smiled.

A learning structure that minimized sensor-based data and pushed camera images directly to the cloud.

This was exactly the direction I wanted to go.

“Sir.”

Kim Se-bin cautiously spoke again.

“Autonomous driving development is proceeding smoothly with support from Dojin Soft, but the Luna Link side is under severe financial pressure.”

Park Cheol-jin seemed to have been waiting for this.

He immediately added an explanation.

“Both the Leviathan and Luna Link projects are black holes sucking up money. Just organizing this quarter’s expenditures……”

Park Cheol-jin pressed the remote to flip the slide.

[Luna Link Funding Execution Status (Provisional)]

(1) Leviathan Infrastructure: $28.4 billion

• Expansion of simultaneous Stage 1 and 2 production lines

• Redesign of engine turbopumps

• Sky Ripper upgrade

• Flight insurance, safety certification costs

• …….

(2) Luna Link Satellites: $33.8 billion

• 3-ton satellite body manufacturing and assembly

• Laser interlink equipment

• Ka-band payload mass production

• 5 ground gateway stations

• Backhaul core network……

(3) Goheung Space Center: $13.2 billion

• Launch pad construction and maintenance

• Environmental test building, vacuum chamber, thermal-vibration equipment purchases

• Control room GNSS equipment……

“$75 billion was spent just this quarter?”

“……Yes.”

I knew it was a costly project.

But it was definitely at a level where the word “serious” was appropriate.

“You said most of the $120 billion raised during the IPO was also invested in the Luna Link project, right?”

Not only that, funds secured from Brexit investments and profits from the construction and defense sectors were also mostly funneled into this.

Executive Vice President Oh Seong-hak spoke with a grave expression.

“Right now, we’re holding on with money earned by Dojin Tech and Dojin Soft, but we can’t keep doing that forever.”

“That’s right. Now is the time for the Luna Link side to establish its own revenue model.”

To complete the Luna Link project, we would need to launch at least 5,000 and up to 40,000 additional satellites.

The problem was that I couldn’t cover all that with my own money.

No matter how reusable the launch vehicle was, an enormous amount of funding was required for each Leviathan launch.

“As you said, sir, each Leviathan launch costs $80 to $100 million. While reusable, the first stage can’t be reused indefinitely.”

“You said it’s difficult to use it more than five times due to stability concerns, right?”

Director Kim Se-bin nodded.

“If pushed, it might be possible to use it more than ten times, but if an explosion occurs during that process, all the satellites onboard could be lost.”

Considering Leviathan’s reusability lifespan, after five uses, we would ultimately need to build a new launch vehicle at a cost of $400 million.

Therefore, to cover the costs, we absolutely had to secure commercial launch contracts like the “lunar probe bidding” mentioned in the feed.