YT YouTube2Text

YouTube2Text → Transcripts → El chip imposible: La barrera del 1 nanómetro acaba de caer

El chip imposible: La barrera del 1 nanómetro acaba de caer — Transcript

by NovaGea · 2,735 words · 454 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:01The semiconductor industry operates on
0:04a scale we simply cannot fully
0:06comprehend, much less explore
0:08exhaustively. But within this vast
0:11structure lie individual discoveries
0:14that alone challenge our understanding
0:16of what it means to advance a
0:18civilization. Advancements so decisive
0:22that they make everything before them
0:23pale into insignificance. And deep
0:26within Chinese semiconductor
0:28laboratories, we believe we have found
0:30the most disruptive one yet. But then,
0:33how far can the impact of a single
0:36material reach? And there will be even
0:39greater barriers waiting to be broken
0:41down. I'm Rafael González and you are
0:46watching Novagea. Join me as we climb
0:49into the realm of impossible chips.
0:51From the materials that defined an era
0:54to the discoveries that, unbeknownst to
0:56almost everyone, have just changed the
0:58technological game forever. Think about
1:04the mobile phone you carry in your
1:05pocket. It has more computing power
1:08than the enormous computers that took
1:09humans to the moon. But this everyday
1:12miracle has its days numbered. For over
1:1550 years, engineers have managed to
1:17double the number of transistors on a
1:20chip by making them smaller and smaller
1:22. A transistor is basically a
1:25microscopic switch that can let
1:27electricity pass or block it to create
1:29the zeros and ones of binary code. This
1:33progression has a famous name: Moore's
1:35Law. But when shrinking silicon pieces
1:39to sizes of two or three nanometers, a
1:41disaster occurs at the subatomic level.
1:44Electrons begin to behave chaotically.
1:47They escape through the walls of the
1:49switch in a phenomenon called the
1:50quantum tunneling effect. The chip
1:53becomes unstable, consumes huge amounts
1:55of energy, and heats up to the point of
1:58potentially melting. Silicon, the
2:01undisputed king of our era, can no
2:03longer keep up. Graphene was tried, but
2:06it had a fatal flaw. It could not be
2:08turned off easily. A completely flat
2:11material of atomic thickness was needed
2:13, but one that behaved like a true,
2:15controllable semiconductor. And it
2:18turns out that material already existed
2:20, only no one knew how to manufacture
2:22it on an industrial scale, or at least
2:24that was the case until now. Scientists
2:31have known for years that Molybdenum
2:33disulfide, known as MoS2, is one of the
2:35most promising 2D semiconductor
2:37materials on the planet. Structurally,
2:41it is like an atomic sandwich: a thin
2:43layer of molybdenum atoms trapped
2:45between two layers of sulfur atoms. The
2:48result is a sheet that is exactly three
2:51atoms thick. Yes, just three. And that
2:55extreme thinness is not just a curious
2:57fact; it is what gives it unique
2:58properties. Being a two-dimensional
3:02structure, electrons can only travel in
3:04a straight line without bouncing up or
3:06down, because in this material,
3:08physical height does not exist. It’s
3:11like a highway for levitating cars.
3:14Maximum speed, minimum friction. But
3:16there is something even more surprising
3:18. Since the material is purely atomic
3:21in thickness, it is inherently flexible
3:23and almost transparent. We are talking
3:26about processors that can be folded
3:28like a sheet of paper or embedded
3:30directly into glass and windows,
3:32opening the door to electronics that
3:34literally disappear into everyday
3:36objects. Furthermore, this sheet is the
3:39perfect candidate for vertical stacking
3:42. With silicon, manufacturing 3D chips
3:44is a nightmare, because the bottom
3:46layer fries the one above it. With
3:49materials that generate almost no heat,
3:51you can build skyscrapers of
3:53transistors, stacking floors like an
3:55atomic lasagna, multiplying the power
3:57in the same square millimeter. The
4:00problem is that for years nobody knew
4:02how to manufacture this material on an
4:04industrial scale, even though it works
4:06in the lab. Small crystals can be grown
4:10using a process called chemical vapor
4:13deposition, or CVD, where gaseous
4:15materials are injected into a chamber
4:18and left to crystallize on a surface.
4:22The results are clean and of very high
4:24quality, but they are tiny and
4:26inconsistent, which makes them
4:28completely useless for manufacturing
4:30real chips. Processors aren't made in
4:34fragments; they are made on wafers,
4:36large uniform discs that factories
4:38process at scale. So, researchers
4:42turned to a more industrial technique:
4:44metal-organic chemical vapor deposition
4:47, or MOCVD. But when they analyzed the
4:52resulting material, they found defects
4:54everywhere, and worse, carbon
4:56contamination. The metal-organic
4:59compounds used in the process were
5:00decomposing and poisoning the
5:02crystalline lattice. To understand the
5:05real impact of this, we have to talk
5:07about electron mobility. It is the
5:10metric that measures the speed at which
5:11electrons move through a semiconductor
5:13material. The higher it is, the faster
5:16the chip processes information. In lab
5:19samples using CVD, electron mobility
5:22was excellent. In sheets produced with
5:25industrial MOCVD, it was a tiny
5:28fraction of that. The material was
5:30practically useless. The dilemma was
5:33brutal. CVD offered impeccable quality,
5:36but it was impossible to scale. MOCVD
5:40was scalable, but it produced defective
5:42material. No one could bridge that gap
5:45until a team in Nanjing found the way.
5:48Because sometimes the solution to an
5:51engineering problem isn't to force the
5:53process harder, but to find a shortcut
5:55that no one had seen before. In January
6:01of this year, Yin Langan Wuang’s team
6:03at Southeast University in Nanjing
6:05published something in Science magazine
6:07that the semiconductor world had been
6:09waiting over a decade for. Not a
6:12concept, not a simulation, not a
6:14promising preliminary result. A real,
6:17physical 6-inch wafer of a 2D material
6:20that grew quickly, cleanly, and was
6:22ready for an actual production line.
6:26The method is called Oxi-MOCVD. What
6:30Wang and his team discovered was that a
6:32specific step in the crystallization
6:34process acted as a monumental
6:36bottleneck. It required such a huge
6:39amount of energy that it was limiting
6:41both the crystal growth speed and the
6:44quality of the final product. That
6:47single step was responsible for MOCVD
6:49sheets coming out full of defects and
6:51contamination. Their solution was
6:55surprisingly elegant. Introduce oxygen
6:58into the reaction. Oxygen creates an
7:01alternative chemical route, a
7:03lower-energy path that completely
7:05avoids the problematic step, like
7:07finding a highway when you’ve been
7:09stuck on a mountain road for hours. And
7:12here is another technical detail worth
7:14mentioning. To get electricity in and
7:17out of these ultra-thin sheets without
7:19it getting stuck, engineers needed to
7:21solve the so-called contact problem.
7:26Soldering conventional metal wires to a
7:28sheet three atoms thick caused
7:29electricity to bottleneck, as if you
7:31were trying to pass the full flow of a
7:33fire hose through a straw. The solution
7:38was to use alloys of heavy elements
7:40like bismuth, which manage to couple to
7:42the atomic structure of MoS2 without
7:43damaging it, allowing energy to flow
7:45without resistance. The overall results
7:49of Oxi-MOCVD were extraordinary. The
7:52growth speed was more than 100 times
7:54higher than that of conventional MOCVD.
7:58Not 10%faster, not double, 100 times.
8:02The sheets came out with no trace of
8:04carbon impurities, and transistors made
8:07from those wafers showed electronic
8:09mobility more than 10 times higher than
8:11devices built with standard MOCVD
8:13material. The paper itself, published
8:18in Science on January 29, 2026, stated
8:21it bluntly. This method bridges the gap
8:25between the quality of lab-scale CVD
8:27and the industrial scalability of MOCVD
8:30. That sentence condenses decades of
8:33frustration and years of work. The gap
8:37that had kept 2D semiconductors in the
8:39realm of theory for nearly a decade has
8:42just closed, and what comes next makes
8:44this discovery seem like only the
8:46beginning. One week after the Oxi-MOCVD
8:53paper was published, a second Chinese
8:55team published their own article in
8:57Science. This one came from the
9:00University of Science and Technology
9:02Beijing. Their breakthrough was
9:04different, but completely complementary
9:06. They had developed a technique that
9:09allows scientists to observe the growth
9:11of MoS2 crystals in real time under a
9:13transmission electron microscope. But
9:18why is this so important? Because
9:22crystal defects, those tiny
9:24imperfections that destroy a device's
9:26performance, form during the growth
9:28process itself. If you can see them
9:31form in real time, you can understand
9:33why they appear. And if you understand
9:36the why, you can design the process to
9:38avoid them. Until now, defect control
9:41in the manufacturing of 2D materials
9:43was largely a trial-and-error process.
9:46You would grow the material, measure
9:49its quality, adjust parameters, and
9:51repeat. A slow, expensive cycle with
9:54little real diagnostic capability. This
9:57new technique turns that artisanal
9:59process into something systematic and
10:02scientific. It's the difference between
10:05guessing where a pipe leak is and
10:07having a camera inside the pipe. Two
10:10publications in the same week. Both in
10:13Science, both from Chinese teams, both
10:16advancing the same technology from
10:18different and complementary angles.
10:21That is no coincidence; that is a field
10:23reaching critical mass. But how many
10:28institutions are behind all this and
10:29how deep does the coordination really
10:31go? Jin Llan Wang's team didn't just
10:37appear out of nowhere. It is part of a
10:39wave of research that has been
10:41systematically dismantling the barriers
10:44to the commercialization of 2D
10:46semiconductors. In 2022, Kaiju's team
10:50at Peking University published a batch
10:53production method for 2D wafers using a
10:55substrate stacking technique, producing
10:58wafers between 5 and 30 cm in diameter.
11:03It was the first credible demonstration
11:05that wafer-scale production of 2D
11:07materials was not just theoretically
11:09possible, it was physically achievable.
11:13Later, in 2025, the same team went
11:15further with a different 2D material,
11:18indium selenide, which has attractive
11:20properties for applications where MoS2
11:23falls short, opening a second parallel
11:26path of development. Also, that same
11:29year, a team from Fudan University in
11:31Shanghai did something that made the
11:34entire field pay attention. They built
11:38a processor, not a transistor, not a
11:40circuit, but a fully functional 32-bit
11:43microprocessor built with 2D
11:45semiconductor materials and a RISC-V
11:47architecture, which they called Wuji.
11:52It is worth pausing here for a moment
11:54on the term RISC. It is a processor
11:57architecture, that is, the set of
11:59instructions that defines how the chip
12:01understands and executes commands. What
12:04makes it special is that it is open
12:06source. Any team in the world can use
12:09it, modify it, and build upon it
12:11without paying licenses or depending on
12:13companies like Intel or ARM. For China,
12:17working with RISC-V is not just a
12:19technical choice; it is a declaration
12:21of technological independence. And Wuji
12:25, with its 5,900 2D MOS transistors
12:27executing 32-bit RISC instructions, is
12:30proof that this independence now has a
12:32physical form. From growing small
12:36crystals in 2020 to a functional
12:38processor in 2025. The timeline is
12:41accelerating. Peking University,
12:45Southeast University in Nanjing, Fudan
12:47University in Shanghai, and the
12:49University of Science and Technology in
12:52Beijing. Four different institutions,
12:56multiple lines of research, all
12:57converging toward the same goal. This
13:01is not a laboratory having a stroke of
13:03luck; it is an ecosystem that has been
13:06built, funded, and coordinated to solve
13:08a specific problem, and they have just
13:10solved it. The Gin One team has a 6-
13:17inch 2D wafer that grows at 100 times
13:19the speed of the previous industrial
13:22method, with no carbon contamination
13:24and 10 times higher electron mobility.
13:28A second team observes crystal growth
13:30in real time. Peking University
13:33produces wafers in batches of up to 30
13:36cm, and Fudan has demonstrated that 2D
13:39materials can run a real processor. The
13:43manufacturing bottleneck, the only
13:45problem that kept all this locked in
13:47the lab, has just received a credible
13:49and scalable solution. And the
13:51implications go far beyond conventional
13:54hardware. Think about this. Processors
13:57will be so tiny and efficient that they
13:59could be integrated into ingestible
14:01medical capsules. Small smart pills
14:05that, once swallowed, navigate your
14:07bloodstream monitoring health markers
14:09in real time, detecting anomalies
14:11before they become diseases, all
14:13without the risk of overheating or
14:15damaging surrounding tissues. Something
14:19that was physically unfeasible with
14:21silicon due to the heat it generated.
14:23The most powerful artificial
14:24intelligence will no longer need the
14:26cloud to operate. It will run natively
14:29on your glasses, your watch, or in
14:31invisible sensors all around you. And
14:35the reduction in energy consumption for
14:37data centers could be one of the key
14:39factors to avoid the energy collapse
14:41that massive digitization threatened to
14:43cause. 2D semiconductors are no longer
14:47the future; they are the present being
14:49manufactured right now. But there is
14:52one more layer to this story, one that
14:54goes beyond science and has direct
14:57consequences for who will control the
14:59world in the coming decades. If you
15:05thought international tension over
15:07silicon microchips was extreme, what
15:08comes next is on another level. For
15:11years, the semiconductor war has
15:13consisted of defending silicon's
15:15strategic bottlenecks. The West
15:19dominates that chain largely thanks to
15:21the extreme ultraviolet lithography
15:23machines from the European company ASML
15:25, without which the world's most
15:27advanced chips cannot be manufactured.
15:30Export restrictions on those machines
15:32to China have been one of the most
15:34potent geopolitical instruments of the
15:36last decade. But if the material
15:39platform changes, all that advantage
15:41evaporates. MoS2 does not depend on
15:44ASML machines and does not need the
15:47same lithography processes. It is not
15:49tied to the same supply chains. It is,
15:53in essence, a completely new playing
15:55field, and China knows it perfectly
15:56well. Suffocated for years by those
16:00blockades, China has not tried to catch
16:02up with the West at its own game. It
16:04has decided to change the game entirely
16:06. The four university teams we have
16:09seen are no coincidence. They are the
16:12visible expression of a coordinated
16:14strategy with massive state funding
16:16aimed at monopolizing the next era of
16:18information processing. Meanwhile, on
16:21the Western side, the response has been
16:23the CHIPS Act. Passed in the United
16:28States in 2022, it is a $ 52 billion
16:30public investment package intended to
16:33reshore semiconductor production to
16:35American soil and fund research into
16:37new technologies. The idea is simple:
16:41reduce dependence on Asia and maintain
16:44a technological edge. Laboratories like
16:48IMEC in Belgium, also supported by
16:49equivalent European funds, work in
16:51parallel seeking to integrate 2D sheets
16:53onto traditional silicon circuits,
16:55trying to extend the lifespan of their
16:57multibillion-dollar factories without
16:59having to abandon what they already
17:01know how to do. It is a legitimate bet,
17:05but it is essentially a conservative
17:07one in the face of a strategy that
17:09points directly to the future. Whoever
17:13wins this race will not just sell more
17:15computers; they will hold the master
17:16key to the new world order in
17:18artificial intelligence, military
17:19technology, and global supply chains.
17:23The problem is that the pieces on the
17:25board have already moved much further
17:27than anyone is willing to admit. We are
17:33facing one of those rare moments in the
17:35history of technology where change does
17:37not arrive gradually, but all at once.
17:40For decades, silicon seemed immovable.
17:43It was the material, the standard, the
17:46foundation upon which everything else
17:48was built, but physics does not
17:50negotiate. And when silicon hit its
17:53limits, the question was not whether
17:55there would be a successor, but when
17:57and who would find it first. The
18:02distance between a lab and a commercial
18:04factory remains real, and the
18:05challenges of large-scale production
18:07have not disappeared, but the
18:09fundamental bottleneck has just
18:10vanished. What remains is engineering,
18:14and engineering with enough investment
18:17and time always finds a way. The era of
18:20the one-nanometer chip has already
18:22begun to beat. So whether it's through
18:27new growth techniques, new materials,
18:30or new processor architectures, or
18:32perhaps even a geopolitical race that
18:35is paradoxically accelerating global
18:37scientific progress, 2D semiconductors
18:40are exploring many ways to mature and
18:42reach the world. And once out of the
18:46labs in Nanjing, the vast distances
18:48that separate pure research from
18:50industrial production require unique
18:52adaptations and unusual flexibility if
18:54they are to take root in a global
18:56economy built on silicon. But what
19:00fascinates me most about all of this is
19:02knowing that what started as a quantum
19:04physics problem is becoming one of the
19:07most important technological and
19:09geopolitical phenomena of our time,
19:11where this subject is no longer just a
19:13scientific curiosity, but it truly
19:15feels like the semiconductor world
19:18could have taken a thousand different
19:20paths, and yet, it is taking this one.
19:23After all, sticking to silicon would
19:25have been accepting a limit. And as we
19:28already know, humans do not accept
19:30limitations. A big hug to the patrons
19:36who support my informational
19:38independence. Subscribe and leave a
19:40like to support the channel. Thanks for
19:43your visit and see you in the future.

About this transcript

This page contains the full transcript of El chip imposible: La barrera del 1 nanómetro acaba de caer by NovaGea, generated from the public captions YouTube serves with the video. The transcript has 2,735 words across 454 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 translate the transcript into 100+ languages.

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, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.