【国策の最深部】ラピダス稼働で覚醒する古河電工。AI半導体バブルの「次」を担う光電融合の正体 — Transcript
Full transcript
- 0:00Concentration of data centers in
- 0:02Hokkaido for Rapidus operations. Many
- 0:06investors here flock only to
- 0:08semiconductors, power, or recently
- 0:10optical fibers, but they are actually
- 0:12overlooking the most fatal bottleneck
- 0:14in AI design. It is not the volume of
- 0:19data communication, but the heat
- 0:21generated inside the servers. Now, no
- 0:24matter how much AI semiconductors
- 0:26evolve or how thick the cables
- 0:28connecting Hokkaido to Tokyo become,
- 0:30there is the massive heat and power
- 0:32consumption generated when converting
- 0:34electrical and optical signals within
- 0:36data centers. Unless this is solved
- 0:39from the roots, Japan's communication
- 0:41infrastructure will physically collapse
- 0:43. The one holding the key to the
- 0:46solution—NTT's national strategy, the
- 0:49IOWN initiative, and its core
- 0:51technology of photonics-electronics
- 0:53convergence—is Furukawa Electric,
- 0:55which we will dive into today. In the
- 0:58second half of this video, we will
- 1:00compare the decisive weaknesses of
- 1:02Furukawa Electric with its
- 1:03market-dominating rival Sumitomo
- 1:05Electric, and examine the unique
- 1:07strengths that more than compensate for
- 1:09them, using specific figures and charts
- 1:11. By watching until the end, you should
- 1:14gain the perspective to logically
- 1:16identify the winners of the industrial
- 1:18structure years from now, without being
- 1:20swayed by superficial AI-related
- 1:22earnings figures. Why is this major
- 1:25wire and cable company the next
- 1:27protagonist after the AI semiconductor
- 1:29bubble? Today, we will unravel the
- 1:31heart of this national infrastructure.
- 1:34Hello, everyone. This is the Kabu no Ki
- 1:36Channel. Please subscribe and turn on
- 1:40notifications now so you don't miss the
- 1:42next important structural changes or
- 1:44the latest analytical updates. Let's
- 1:49first organize the background on why we
- 1:51should view Furukawa Electric not just
- 1:53as a thematic stock, but from the
- 1:55broader perspective of structural
- 1:57changes in communication infrastructure
- 1:59. Currently in Hokkaido, Rapidus is
- 2:03building a state-of-the-art
- 2:05semiconductor plant in Chitose, and
- 2:07Sakura Internet is launching one huge
- 2:09national-project data center after
- 2:11another in Ishikari. Hokkaido is now
- 2:14becoming the central hub for computing
- 2:17resources in Japan and Asia. Why
- 2:20Hokkaido instead of Tokyo? It is
- 2:25because only Hokkaido has the vast land
- 2:27, cool climate to cool servers, and
- 2:29massive power required for generative
- 2:31AI processing. In Tokyo and Osaka,
- 2:36power grid capacity has already reached
- 2:39its limit, making it physically
- 2:41difficult to even build new data
- 2:43centers. The next challenge that arises
- 2:46from this is distance. Massive amounts
- 2:49of AI data processed in Hokkaido must
- 2:52be sent instantaneously and without
- 2:54delay to Tokyo, the largest demand hub,
- 2:57a straight-line distance of over 800km.
- 3:00The installation of terrestrial and
- 3:02undersea fiber-optic networks to
- 3:04connect these points is currently
- 3:06proceeding at a rapid pace. However,
- 3:09there is a major bottleneck here that
- 3:11the market is overlooking. Many people
- 3:13think like this: "In that case, we
- 3:15should just lay down plenty of thick,
- 3:17high-performance cables that can carry
- 3:19more data at once.""Therefore, major
- 3:22optical cable companies must be
- 3:23profitable." Certainly, that
- 3:25perspective is not wrong. But the core
- 3:28problem is not inside the cables; it is
- 3:30happening inside the data centers,
- 3:32specifically at the server entry and
- 3:34exit points. Currently, light signals
- 3:38traveling at the speed of light through
- 3:40fiber optics are converted into
- 3:42electrical signals just before entering
- 3:44the data center servers. This is
- 3:46because current semiconductors, such as
- 3:49GPUs and CPUs, can only perform
- 3:51calculations using electricity. This
- 3:54light-to-electricity conversion process
- 3:56, known technically as photoelectric
- 3:58conversion, actually generates an
- 4:00incredible amount of heat. Currently,
- 4:03communication speeds are accelerating
- 4:06by orders of magnitude to 400G, 800G,
- 4:08and beyond. As the volume of data
- 4:12explodes, the conversion work at the
- 4:14gateway becomes intense, and heat
- 4:16generation skyrockets. There is even
- 4:19data suggesting that 30%to 40%of the
- 4:21enormous power consumed by current data
- 4:24centers is taken up not by server
- 4:26calculations themselves, but by air
- 4:28conditioning to cool the generated heat
- 4:30. In other words, no matter how many
- 4:33high-performance cables you lay between
- 4:36Hokkaido and Tokyo to send massive
- 4:38amounts of data, a massive traffic jam
- 4:40caused by heat is occurring at the data
- 4:42center entrance. If servers go down due
- 4:45to heat, they cannot function as
- 4:47infrastructure at all. This is the
- 4:49physical limit that current
- 4:50communication infrastructure is facing.
- 4:53As an aside, the technology to simply
- 4:55build high-capacity cables will not
- 4:57fundamentally solve this heat problem.
- 5:00The structure of infrastructure
- 5:02investment money is forced to shift
- 5:04from the cables themselves toward
- 5:05resolving this heat problem. This is a
- 5:09physical bottleneck that simply cannot
- 5:11be overcome. This is the very reason
- 5:13for the existence of NTT's IOWN
- 5:15initiative as a national project, and
- 5:17it ties directly into the scenario
- 5:19where Furukawa Electric, whose
- 5:21technology we will compare in detail
- 5:23later, holds the key to the next
- 5:25generation. I would like to ask you all
- 5:27one question here. As you are all well
- 5:30aware, the recent frenzy in the
- 5:32electric wire sector, especially
- 5:34Fujikura, has created a historic bull
- 5:37market driven by the AI boom in North
- 5:39America. So, even though both are among
- 5:42the world's top-tier electric wire
- 5:44companies, why is the market's initial
- 5:46capital so heavily concentrated in
- 5:48Fujikura's physical cables rather than
- 5:50Furukawa Electric's optical devices?
- 5:52The answer lies hidden not in the
- 5:54superficial technological superiority,
- 5:57but in the cold, hard fact of a
- 5:58divergence in the time horizons that
- 6:00market capital is targeting. Let me
- 6:03state the conclusion first. The market
- 6:05is currently pouring funds into the
- 6:07physical lines that are needed
- 6:09immediately, but it has not yet fully
- 6:10priced in the value of the converters
- 6:12that will solve the heat problems
- 6:14several years down the line. On the
- 6:17screen, I have displayed a comparison
- 6:19summarizing the decisive time-horizon
- 6:21gap between Fujikura and Furukawa
- 6:22Electric, which are currently splitting
- 6:24the telecommunications infrastructure
- 6:26demand. The meaning of this table is
- 6:30very simple. Fujikura's current
- 6:32overwhelming strength comes from its
- 6:34ultra-high-density cable technology,
- 6:36which meets the immediate demand of
- 6:38North American tech giants wanting to
- 6:40pack as many optical fibers as possible
- 6:43into limited underground conduits when
- 6:45building data centers. This is
- 6:47certainly a magnificent technology, and
- 6:49the current rise in their stock price
- 6:51is a justified one backed by their
- 6:53performance. However, as I mentioned
- 6:56earlier, simply packing more cables
- 6:58into the conduits does not solve the
- 7:00fundamental heat problem. The more the
- 7:02volume of data increases, the more the
- 7:04power consumption due to heat
- 7:05generation will skyrocket. This is
- 7:07where the core technology of Furukawa
- 7:10Electric, opto-electronic fusion, comes
- 7:12into play. Furukawa Electric is a
- 7:15global leader in research on technology
- 7:17that directly converts light and
- 7:19electricity on a semiconductor chip,
- 7:21rather than using the large devices
- 7:23traditionally required to convert
- 7:25electrical and optical signals. In
- 7:29NTT’s IOWN initiative, which is a bet
- 7:31on the company's future, there is an
- 7:33ambitious goal to reduce power
- 7:35consumption to one-hundredth of current
- 7:38levels; making this physically possible
- 7:40is the optical device technology
- 7:42centered on high-power CW lasers, where
- 7:44Furukawa Electric boasts world-class
- 7:47expertise. If I must mention a concern
- 7:50here, it is the fact that for many
- 7:52years, Furukawa Electric has been held
- 7:54back by losses in its automotive parts
- 7:56business, such as wire harnesses, and
- 7:58the worsening market for metal
- 8:00materials. "Even if communications are
- 8:02good, the other businesses are failing,
- 8:04so I can't buy the stock." This has
- 8:06been the primary reason why many
- 8:07institutional and individual investors
- 8:10have hesitated to invest in Furukawa
- 8:12Electric until now. However, please
- 8:14look at this data. Furukawa Electric is
- 8:16currently pushing forward with
- 8:17structural reforms without any
- 8:19exceptions. They are carrying out a
- 8:21transformation of their business
- 8:22portfolio—literally shedding blood—
- 8:24by thoroughly downsizing and
- 8:26restructuring the unprofitable
- 8:27automotive parts business that was
- 8:29eating into profits, while
- 8:30concentrating management resources on
- 8:32information communications, optical
- 8:34infrastructure, and energy
- 8:35infrastructure. In other words, the
- 8:37factors that were holding them back are
- 8:39being eliminated. The company is being
- 8:41reborn into a strong, lean structure
- 8:43where the high profit margins from
- 8:45optical-electrical fusion devices
- 8:47directly boost the company’s EPS. To
- 8:51harness the tremendous computing power
- 8:53of next-generation, high-performance
- 8:55semiconductors like those manufactured
- 8:57by Rapidus without having it killed by
- 9:00heat, CPO (Co-Packaged Optics)
- 9:02technology, which converts light and
- 9:04electricity right next to the chip, is
- 9:06absolutely essential. And some of you
- 9:09may have realized this after hearing
- 9:11what I've said so far. In fact, the
- 9:13next-generation semiconductors from
- 9:14Rapidus, which is expanding into
- 9:16Hokkaido, and Furukawa Electric’s
- 9:18optical-electrical fusion are
- 9:20completely linked at the foundational
- 9:21level. It is said that due to the
- 9:24evolution of AI, semiconductors will
- 9:26ultimately evolve into a form called
- 9:28CPO. This is a technology that directly
- 9:31attaches optical communication
- 9:33functions—optical transceivers—
- 9:35right next to the computing
- 9:36semiconductor on the same package. It
- 9:39shortens electrical wiring to the limit
- 9:41and transmits data directly from the
- 9:43chip using light. In short, the artery
- 9:46of light connecting Hokkaido to Tokyo
- 9:49begins with NEC's undersea cables,
- 9:51relies on foundations built by Hokkaido
- 9:53Electric Power and Comsys on land, and
- 9:56is ultimately processed within the
- 9:58server at the millimeter scale by
- 10:00Furukawa Electric’s optical fusion
- 10:02devices. This is the true face of the
- 10:04complete supply chain for the
- 10:06infrastructure being promoted as a
- 10:08national project. The market is
- 10:11currently fixated on the immediate AI
- 10:13semiconductor bubble and the
- 10:14extraordinary financial results from
- 10:16Fujikura. However, the true fruits of
- 10:19investment may only be enjoyed by those
- 10:21who place their capital not where the
- 10:23masses are enthralled, but on the
- 10:25resolution of structural bottlenecks
- 10:27silently progressing behind that frenzy
- 10:29. How was that? This time, I have
- 10:32explained the thermal issues facing
- 10:34infrastructure amid the massive
- 10:36national projects of Rapidus operations
- 10:39and data center concentration in
- 10:40Hokkaido, along with the key to solving
- 10:43them: Furukawa Electric’s optical
- 10:45fusion. The biggest mistake to avoid in
- 10:48investing is jumping on the bandwagon
- 10:50at the peak of a theme that everyone
- 10:52already knows. Fujikura's remarkable
- 10:56success deserves proper recognition,
- 10:58but what we should focus on now is the
- 11:00quiet turning point where the
- 11:02infrastructure investment phase shifts
- 11:04from laying physical lines to devices
- 11:06that manage server heat. As a near-term
- 11:11event, I want you all to keep an eye on
- 11:13the next quarterly earnings
- 11:15announcement. What we should track here
- 11:18is not the superficial scale of revenue
- 11:20. Whether the losses from drag-on
- 11:23businesses like automotive parts are
- 11:25shrinking as planned, and whether the
- 11:27profit margins in the information and
- 11:29communications division for optical
- 11:31devices are trending upward. The moment
- 11:35these two figures cross is when
- 11:37Furukawa Electric’s EPS will see a
- 11:39dramatic leap. While much of the market
- 11:43is captivated by short-term AI hype and
- 11:45easy-to-spot stock surges, you have now
- 11:47grasped the inevitable physical
- 11:49bottlenecks arriving in a few years and
- 11:51the structural flow of capital toward
- 11:54resolving them. Do you think current
- 11:57investment funds should be weighted
- 11:59more toward Fujikura, with its visible
- 12:02wiring, or Furukawa Electric, with its
- 12:04devices that solve future heat issues?
- 12:07Please let me know your thoughts on
- 12:09portfolio strategies or any hidden
- 12:11infrastructure stocks that you think
- 12:13could skyrocket in the comments section
- 12:14. And finally, I would like to announce
- 12:18that memberships have been reopened. I
- 12:21have opened memberships on this channel
- 12:23to ensure I can continue independent
- 12:25research and analysis without being
- 12:27influenced by platform trends.
- 12:30Currently, this is not a place for
- 12:31trading specific investment information
- 12:34, but is operated as a venue for purely
- 12:36supporting those who agree with the
- 12:38principles of my activities. I would
- 12:40like to take this opportunity to
- 12:42express my deepest gratitude to
- 12:43everyone who has already signed up as a
- 12:46new member upon this reopening. First,
- 12:48I aim to solidify the foundation of
- 12:50these activities and continue to
- 12:52explore the structure of Japanese
- 12:54stocks from a mid-to-long-term
- 12:55perspective together with all of you.
- 12:58If you agree with the purpose of these
- 13:00activities, please consider joining. I
- 13:04will see you all in the next video.
- 13:07Thank you for watching until the very
- 13:09end.
About this transcript
This page contains the full transcript of 【国策の最深部】ラピダス稼働で覚醒する古河電工。AI半導体バブルの「次」を担う光電融合の正体 by 株の気づきチャンネル, generated from the public captions YouTube serves with the video. The transcript has 1,962 words across 350 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.