When Copper is Blocked and Light is Too Expensive: The Choice of a Korean Startup Backed by NVIDIA

The Bottleneck in the Wires

In 2025 alone, the world’s largest cloud companies poured hundreds of trillions of won in capital expenditure into data centers.

A single NVIDIA H100 chip costs over 40 million KRW. When thousands of these are bundled into a cluster, the cost quickly balloons into the hundreds of billions of won. People assume most of this money goes into the compute chips.

They are wrong. The most expensive part isn’t the chip; it’s the time the chips spend waiting for each other.

When thousands of GPUs train a single neural network, they must exchange intermediate calculation results at every step. If one unit finishes its calculation but doesn’t receive a synchronization signal from others, it cannot move to the next stage. The industry calls this idle state a ‘Stall.’ And stalls don’t happen in the chips—they happen in the wires.

Specifically, it happens in moments like this:

At the next-generation 1.6 Tbps transmission standard, the effective transmission distance of copper cables collapses to under 1 meter due to the laws of physics. However, the average server rack height exceeds 2.5 meters. We have reached a point where it is physically impossible to connect the GPU at the very bottom of a rack to the network switch at the top. The industry calls this phenomenon the ‘Copper Cliff.’

Optical cables, proposed as an alternative to this cliff, solve the distance issue but require 15–20W of power per port. For hyperscale data centers running tens of thousands of ports, this translates into massive electricity bills and becomes a primary driver of carbon emissions.

Copper is too short, and light is too expensive. In the gap created by this dichotomy, plastic has emerged.

The First Wall: Copper Swallows High Frequencies

The Physics of the Skin Effect

Why do copper cables suddenly become useless? The answer lies not in materials science, but in electromagnetics.

When alternating current flows through a conductor, as the frequency increases, the charge is pushed toward the surface rather than the center of the conductor. This Skin Depth (δ) phenomenon is inversely proportional to the square root of the frequency f.

δ = √(ρ / πfμ₀μᵣ)

In plain terms, when the frequency quadruples, the effective cross-sectional area through which current can flow is halved. As the cross-sectional area decreases, resistance increases, and as resistance increases, the signal dissipates as heat. In the ultra-high frequency band of 100GHz and above, a copper conductor acts less like a signal transmission medium and more like a heater incinerating the signal. This is why passive copper cables cannot exceed a distance of 1 meter in a 1.6 Tbps transmission environment, and this distance will shrink further in the 3.2 Tbps generation.

A World After the ‘Copper Cliff’

Field engineers have been coping with this limit by using thicker cables. Thicker wires reduce resistance, but the cables become heavy and rigid, clogging the back of server racks. If exhaust vents are blocked, cooling efficiency drops and internal temperatures exceed critical thresholds. The weight and stiffness of the cables can even physically damage port connectors.

An emergency fix for this problem is the Active Electrical Cable (AEC). By attaching a signal regeneration chip (retimer) to the end of the cable, the reach is extended to 2–2.5 meters. However, that is the limit. It is still insufficient to meet the wiring requirements of 1.6 Tbps generation data centers.

The evolution of copper cables has hit the ceiling of physical laws. Faster cables or thinner copper cables are no longer possible. All that remains is higher frequency and the resulting ever-narrowing effective cross-sectional area of the conductor.

The Second Wall: Light is Expensive

The Cost of Optical Cables

Optical fibers transmit signals through light refraction rather than electromagnetic current, making them completely immune to the skin effect that plagues copper. They can easily cover distances in the kilometer range. So, why don’t hyperscale data centers switch entirely to optical cables?

To convert electrical signals into light, you need hardware. A laser diode emits light, a driver modulates it, an optical DSP restores the signal, and a transimpedance amplifier (TIA) at the receiving end converts it back into voltage. This combination of components is an optical transceiver module.

A single 800 Gbps-class optical transceiver consumes an average of 15–20W. In a massive data center with tens of thousands of ports running simultaneously, the power consumed by the interconnect layer alone is equivalent to a massive power plant.

Cost is also a problem. Optical transceivers require indium phosphide (InP) or gallium arsenide (GaAs) compound semiconductors and silicon photonics packaging, resulting in a unit price 4–10 times higher than equivalent passive copper cables.

Reliability is another weakness. Laser diodes age quickly in high-temperature, high-humidity environments, and thermal shock near server rack exhausts further reduces their lifespan. In distributed parallel computing, a single connection failure can stop an entire training session, forcing a restart from the previous checkpoint. Situations where training that had been running for days across thousands of GPUs is wiped out by the failure of a single optical transceiver’s laser really do happen.

The Limits of Co-Packaging

Co-Packaged Optics (CPO) is an attempt to solve this by placing chips and optical elements in the same package. It minimizes the distance electrical signals travel on the board to improve power efficiency. However, CPO drastically increases the complexity and cost of the manufacturing process. Integrating silicon logic and optical elements into one package is a challenge in semiconductor manufacturing, and above all, it fails to eliminate the fundamental energy inefficiency of optical-to-electrical conversion.

The Third Material: Plastic Swallows Millimeter Waves

In 2014, an idea from a KAIST research lab attacked this dichotomy head-on. It uses no copper and no lasers. The concept was to fire electromagnetic waves into a plastic tube.

Point2 Technology developed the ’e-Tube,’ a system that confines and transmits RF signals in the millimeter-wave (100–260 GHz) band within a specially molded, low-loss polymer resin dielectric waveguide.

Because there is no conductor, the skin effect does not occur at the source. Because there is no laser, there are no optical-to-electrical conversion losses and no risk of laser aging.

What Happens Inside the Waveguide

While the physical principles of dielectric waveguides are similar to optical fibers, the signals they confine are different. While optical fibers confine light at visible light frequencies (hundreds of THz), e-Tube confines electromagnetic waves at millimeter-wave frequencies (hundreds of GHz).

The principle is Total Internal Reflection. A plastic core with a high refractive index is wrapped in a cladding with a low refractive index. The injected electromagnetic waves cannot cross the boundary and propagate along the core. In this structure, signal attenuation is determined solely by the dielectric loss tangent (tan δ) and permittivity (ε) of the plastic material, not by high-frequency metal resistance.

α_dielectric ∝ f · √ε · tan δ

The key to this relationship is that, unlike copper, the rate at which attenuation increases as frequency rises is much more gradual than in metal. This provides the physical basis for signals to travel stably for 10–20 meters even in the ultra-high frequency bands required by the 1.6 Tbps and 3.2 Tbps generations.

From TX Chip to Waveguide: The Role of the MWT Antenna

Computer systems operate on baseband digital signals. If you inject these signals directly into a dielectric waveguide, low-frequency components leak through the boundaries and fail to propagate.

That is why Point2 Technology embedded a proprietary RFIC chipset inside the plug at the end of the cable.

When the transmit chip (TX SoC) receives a baseband digital signal and upconverts it to a 100–260 GHz millimeter-wave analog RF signal, a ‘Microstrip-to-Waveguide Transition (MWT) antenna’ etched onto the chip surface fires the signal into the e-Tube.

MWT matches the planar electrical impedance of the silicon substrate to the three-dimensional electromagnetic impedance inside the dielectric, ensuring electromagnetic waves are directed into the center of the waveguide core without energy reflection or external leakage.

At the other end, an on-chip antenna on the receive chip (RX SoC) captures the arriving electromagnetic waves and uses a downconversion mixer to restore them to the original digital bus signal. The host ASIC, such as a GPU or switch, does not even need to be aware that an e-Tube is in the middle.

The 80-Picosecond Figure

This analog conversion path results in a latency of 80 picoseconds (ps). Optical cable systems using optical transceivers generate microsecond (μs) level latency due to electrical-to-optical conversion, digital signal processing, and optical-to-electrical conversion. e-Tube is 1,000 times faster than that.

In distributed parallel computing, latency is not just a simple speed indicator. When thousands of GPUs perform synchronization calculations, the slowest link determines the processing speed of the entire cluster. This is why 80ps of latency carries 1,000 times the significance.

Comparison Table by the Numbers

When you line up five different interconnect technologies, the picture becomes clear.

e-Tube fills the gap between the distance abandoned by copper and the cost that optics cannot handle. Designing this technology to handle rack-level server wiring in the 10–20 meter range is its goal.

The Company’s Roots and Two Cash Cows

Point2 Technology was founded in March 2014. It was co-founded by CEO Sean Park, who led high-speed communication system chip design for over 13 years at Marvell and TeraSquare, and Professor Hyunmin Bae (currently the head of the KAIST Startup Institute) along with his Ph.D. research team. The company operates with a dual structure, with its headquarters in San Jose, California, and core R&D handled by its Korean subsidiary.

It took time for e-Tube to establish itself in the market. Two business lines fill that gap.

25G Range Extender: The Hidden Bottleneck in 5G Base Station Networks

In 5G networks, the ‘Fronthaul’ optical links connecting base stations to core networks are being upgraded from 10 Gbps to 25 Gbps. This creates an unexpected problem.

‘Chromatic Dispersion,’ where the refractive index of light varies with wavelength in optical fiber, distorts signals in proportion to the square of the frequency ratio. An optical line that easily covered 70–80km at 10 Gbps sees its effective distance collapse vertically to 15km when upgraded to 25 Gbps. It is a paradox where faster transmission results in shorter reach.

The traditional solution was to install massive, expensive external dispersion compensation filter devices at every relay point and have technicians manually tune them to match the line length. Point2 Technology’s range extender solves this with a semiconductor chip the size of a fingertip inside the optical transceiver module. An Electronic Dispersion Compensation (EDC) engine restores the completely collapsed eye pattern in real-time, returning the effective transmission distance to 60–70km. No external devices or field tuning required. You just replace the optical transceiver once.

Based on this technology, Point2 Technology signed a long-term development partnership with Sumitomo Electric Industries of Japan and secured a contract for the mass production of core ICs for 25G optical transceivers. This is currently the company’s primary cash flow.

Smart Retimer P1B121: Half the Power, One-Twentieth the Latency

The second cash cow is the 112G PAM4 smart retimer chipset ‘P1B121’ for data centers. A retimer is a chip that cleans up distorted signals at the end of high-speed electrical cables; it is a required component for next-generation 800 Gbps and 1.6 Tbps Active Electrical Cables (AEC). The problem is that competitors’ products consume too much power.

The P1B121 achieves a 3.0W operating power, about 50% lower than comparable competitors. The latency is also under 3 nanoseconds (ns), one-twentieth of the industry average. This is the result of mixed-signal optimized design that maximizes analog signal path processing. 3W and 3ns—these two numbers mean more than just digits. They prevent retimer latency from becoming a bottleneck in large-scale accelerator clusters that require ultra-low latency synchronization, while simultaneously creating room in the power budget.

Why NVIDIA Invested Directly in Korea

In 2024, an unusual event occurred in the global semiconductor market. NVIDIA’s strategic venture investment arm, NVentures, made its first-ever direct equity investment in a Korea-based semiconductor startup. The target was Point2 Technology.

NVentures is an organization that executes strategic investments to strengthen NVIDIA’s future platforms and technology ecosystem, rather than seeking purely financial returns. This move can be interpreted as an intention to incorporate the e-Tube transmission layout as a core backbone into the scale-up fabric of next-generation accelerator architectures following Blackwell.

Including this investment, Point2 Technology has completed a total of $76 million (approximately 112.1 billion KRW) in Series B funding. Maverick Silicon led the round, with UMC Capital, Bosch Ventures, and Molex participating. The list of investors suggests a strategic ecosystem building rather than simple financial support.

The participation of the global automotive parts giant Bosch hints at the possibility of e-Tube entering the automotive electronics market. The investment from wire manufacturer Molex is a case where production partnerships have evolved into capital relationships. The 2026 BloombergNEF Pioneers award is external validation of this technology.

The Fabless Paradox: How to Dominate the Global Supply Chain Without a Factory

Point2 Technology does not own manufacturing plants. That is the strategy.

They do not manufacture finished e-Tube cables themselves. Instead, they design the RFIC chipsets that control electromagnetic modulation and license the geometry and bonding design technology of the e-Tube waveguides. Actual production is handled by the world’s largest connector manufacturers.

Molex, Foxconn Interconnect Technology (FIT), and Amphenol. These three companies are proven vendors that have already supplied trillions of won worth of volume to the data center supply chains of Amazon AWS, Google Cloud, Microsoft Azure, and Meta.

For Point2 Technology to enter the hyperscaler supply chain directly, it would have to pass years of rigorous factory reliability audits (Supplier Audit). However, by shipping finished ARC (Active RF Cable) products embedded with e-Tube chipsets through the brands and logistics networks of already-registered partners, they can shorten this process.

They only design the chips. Manufacturing is done by partners who have already built trust. They also use the partners’ channels. This is the secret to why the fabless model generates high operating profit margins.

---

Shifting Revenue Mix Toward 2030

Currently, Point2 Technology’s cash comes from 25G range extenders and smart retimers. This is the fuel that sustains the company today. However, the direction the company has mapped out points elsewhere. The core of their internal roadmap is to transition 90% of total revenue to e-Tube-based products by 2030.

There is a market structure supporting this transition. Between 2026 and 2028, global data centers will fully transition to the 1.6 Tbps and 3.2 Tbps generations. At this point, copper cables will inevitably be replaced entirely by ARC or optical cables due to physical limitations. This is the size of the replacement market e-Tube is targeting.

Beyond the external cable stage, deeper integration awaits. NPE (Near Package e-Tube) places ultra-thin e-Tube waveguides right next to the silicon interface inside the accelerator board. CPE (Co-Packaged e-Tube) is a structure that integrates RFIC elements directly inside the GPU packaging substrate. At this stage, copper wiring noise on the board disappears, and electromagnetic waves are fired directly from the accelerator chip into the dielectric tube.

Information that they have signed a confidential MOU with at least one of the world’s top GPU manufacturers and are preparing a prototype demo for the second half of 2026 suggests that this roadmap is not just a declaration, but in the execution stage.

E-Tube’s penetration path does not stop at data centers. From terabit-level signal wiring inside ultra-thin premium display panels to noise-canceling wiring between autonomous driving ECUs and multi-channel sensors, the physical principles of e-Tube can be applied anywhere electromagnetic interference is a problem and space is limited.

KOSDAQ IPO, Where Are They Really?

When I first wrote this, Point2 Technology’s goal was to file for a KOSDAQ technology special listing preliminary review in the second half of 2026. However, checking again a few months later, that schedule has not been maintained.

According to the most recent reports, the company’s IPO target has been pushed back to “as early as next year, as late as the year after.” That is, between 2027 and 2028. Because the headquarters is a U.S. entity, there is even talk of shifting direction toward global capital markets like NASDAQ instead of KOSDAQ. The underwriters remain the joint system of Samsung Securities and Korea Investment & Securities, and the target corporate value at listing of over 1 trillion KRW remains unchanged.

The background to the delayed schedule is not much different from what was initially known. The cross-border governance structure between the U.S. San Jose headquarters and the Korean subsidiary has been a hurdle. Coordinating the U.S. federal law requirements for for-profit corporate governance with the Korea Securities Depository and Korea Exchange’s technology special listing lock-up requirements and cross-border equity settlement structures has required a longer period of coordination than expected.

The funds raised in the pre-IPO round just before the listing are planned to be prioritized for foundry initial mass wafer production deposits and the expansion of local U.S. engineering support infrastructure. The IPO schedule itself shows that even if technology is verified, the listing clock does not automatically speed up.

Wires Make Technology Invisible

There is something to pause and think about here.

The problem e-Tube solves is interesting in that it works properly only when the technology does not reveal itself. The best interconnect solution has no presence. GPUs don’t care what the cable is made of. The host system doesn’t even perceive the fact that electromagnetic waves passed through plastic in the middle.

This is the essence of infrastructure. Infrastructure works best when it is invisible.

However, there are moments when infrastructure suddenly becomes visible. It’s when the wires are blocked, the power budget is exceeded, or latency holds back the entire cluster. The data center industry is standing exactly at that point right now. Copper is blocked, optics are too expensive, and AI keeps demanding faster performance.

The moment infrastructure is revealed as a crisis is the condition for new materials to find a niche. The fact that a plastic dielectric has created a third space between copper and optical fiber is not just a story about a better cable being invented. It is a matter of re-engineering the physical constraints of AI infrastructure.

Stories about AI changing the world are common. Stories about what material the wires running that AI are made of are rare. But if a single wire determines the actual performance of a multi-trillion won cluster, the question about the company making that wire is no different from the question about the future of AI infrastructure.