- Where augmented reality goes from here…
- NVIDIA’s strategic metaverse move…
- NFT artist Beeple is back at it again…
One of the most interesting developments in the semiconductor industry this year was announced just a few weeks ago in one of my favorite places on Earth – Japan.
It is an unlikely place for such an announcement. Japan has consistently fallen behind the U.S., Taiwan, and South Korea in semiconductors for the last two decades. The country simply hasn’t invested enough, hasn’t innovated, and suffers from very high costs of manufacturing and a lethargic business culture.
That’s why the announcement of a new $7 billion semiconductor manufacturing plant in Japan was unusual.
Taiwan Semiconductor Manufacturing (TSMC) – a longtime favorite of mine – and Sony teamed up to build a plant in Kumamoto, Japan, that’s scheduled for mass production by 2024.
Kumamoto is on the southern island of Kyushu sitting on the coastline of the Ariake sea. Kyushu is beautiful. I always liked my trips there. The people are nice, the food is great, and things move at a much slower pace than the hurried streets of Tokyo.
Sony is putting up $500 million in the venture, which is worth about 20%. And the real muscle is obviously coming from TSMC.
The focus of the plant will be to manufacture semiconductors using the 22 and 28 nanometer (nm) manufacturing processes. The “nanometer” refers to how far apart transistors are from one another on a semiconductor. The smaller the number, the more advanced the chip. For additional context, 22 and 28 nm process nodes are typically thought of as legacy processes.
For perspective, 22 nm technology was leading edge back in 2012, almost a decade ago. 28 nm came a couple of years earlier.
Since then, the industry has progressed through 14 nm, then 10 nm, 7 nm, and now the latest iPhones incorporate key semiconductors manufactured at just 5 nm. And TSMC is already producing early prototypes of 3 nm products.
So why would TSMC want to invest billions in a manufacturing plant that will produce semiconductors using manufacturing technology that is six or seven generations behind? That’s a decade or more in the semiconductor space.
This is an interesting question, and one that is not well understood.
The reality is that the major semiconductor shortages aren’t happening on the leading edge. They are happening at those older processes like 22 nm and 28 nm. This is where the industry underinvested and did not anticipate the increased demand.
The reality in electronics is that not every device needs bleeding edge semiconductors. Smartphones need them because they are very sensitive to size and power consumption. This is where smaller and smaller dimensions are so useful.
But cars, refrigerators, washing machines, etc. have a lot of space to work with and don’t have the power constraints of a smartphone. Large, less efficient chips are okay and have the added benefit of being less expensive to manufacture.
So the industry needs more capacity from these “older” chips, hence the new plant. But something else is happening behind the scenes. Why not just build the plant in Taiwan where the cost structure is so much better than Japan?
The reality is that TSMC is in a mad scramble to diversify its manufacturing base outside of Taiwan as quickly as it can. It’s the same reason that it announced a major new manufacturing plant will be built in Arizona. We should expect more of the same to come.
Taiwan is under the threat of being taken over by China. There has never been a riper time to do so. To say that the current U.S. administration is weak on China would be a gross understatement. Politics aside, if there were ever a time for China to “strike,” it’s now.
Whether it happens or not, the threat is real. I’m sure that it would be called something innocuous sounding like “administrative control” or the “PROC Office of Oversight.”
But it will certainly be as material as what we’ve seen unfold in Hong Kong over the last two decades… Only I believe that the control will happen far more abruptly.
TSMC knows it. And that’s why it’s taking action. By further decentralizing its manufacturing base, it diversifies risk and reduces the leverage that China would have in using TSMC as a pawn in a geopolitical battle.
Smart move… Nothing but respect.
I only wish they would have started doing this this five years ago so that those new manufacturing plants were already online…
The next big thing in augmented reality…
I’m on the record saying that 2022 will be the year when we start to see the wide adoption of consumer-targeted augmented reality (AR) eyewear. That’s because several big names are set to release AR glasses next year.
As we have discussed before, these glasses will enable apps and games that overlay graphics on our field of vision. And they will tether to our smartphones to leverage their wireless connectivity and computing power. That helps keep the form factor small, as the glasses will need fewer electronics inside.
We can think of these upcoming AR glasses as the first generation of consumer AR eyewear. And that raises the question… What comes next? What will make AR even more seamless?
The answer is AR-enabled contacts. I’m seeing a lot of investment pick up around this.
In the past, we’ve had a look at some “smart” contact lens technology, like the kind being worked on by Samsung. Smart contact lenses, like smart glasses, are an early form of augmentation with limited functionality. We can think of this as a stepping-stone.
But full augmentation is the goal. That’s the technology that is compelling enough for mass market adoption. And one company making good progress is an early stage company called Mojo Vision. We originally talked about Mojo Vision back in April.
Mojo Vision is developing contact lenses that can project the digital world directly into our retinas. This enables high-resolution images that are far more vivid than those that can be displayed on a screen.
Mojo’s Contact Lens
Source: Mojo Vision
In fact, Mojo Vision’s AR contacts can power images with 14,000 pixels per inch. That’s about 300 times as many pixels per inch as we have on our smartphone screens.
This is possible because the contacts focus light onto a very small portion of the retina. And in addition to the enhanced visuals, this approach requires far less power than AR glasses will.
Mojo is also developing its own technology to tether to a smartphone for connectivity. The technology will require even less power than Bluetooth in order to conserve energy stored in tiny batteries contained within the lens.
The attractive point about the lenses is that they will be the least intrusive way to have augmented vision. The added convenience of not having to wear glasses will certainly be attractive to many. And, of course, these lenses are completely hands-free.
AR-enabled contacts also have the benefit of aiding people who have bad vision. They can enhance images and allow wearers to zoom in on objects, just like we do when we zoom in with our smartphone cameras. That’s quite a compelling feature.
That said, Mojo Vision is still very early on in the development process. We are still a few years away from AR contacts.
The first-generation AR glasses will come first. That will be a big investment trend we’ll follow over the coming year. If you’re interested in learning more about that, simply go right here for more info.
NVIDIA’s latest development will power the metaverse forward…
NVIDIA just made yet another brilliant move. The company announced what it calls the NVIDIA Omniverse. It’s a platform designed to accelerate the development of metaverse technology.
Remember, NVIDIA is a semiconductor company. Its core business is hardware. But one of the things NVIDIA has done remarkably well is design software optimized to run on its semiconductor technology. This is NVIDIA’s strategy to empower developers to create applications that use its semiconductor technology.
It did this recently for the autonomous driving field. And now NVIDIA is taking the same approach to the metaverse.
NVIDIA Omniverse rolled out first to 70,000 software developers for testing. The platform incorporates artificial intelligence (AI) to help developers create avatars, landscapes, and virtual objects for any kind of metaverse.
What’s more, NVIDIA Omniverse enables the development of synthetic data. These are computer-generated data sets used to train the AI. This helps create virtual characters that people can interact with in the metaverse, just like they would another person.
In a recent presentation, NVIDIA CEO Jen-Hsun “Jensen” Huang demonstrated this avatar technology by creating an avatar of himself.
It incorporates a form of AI called natural language processing that is used for understanding speech and speech synthesis. This technology can be used as a representation of a person and even speak in the same voice.
NVIDIA CEO Huang and His Omniverse Avatar
Now that the testing phase is complete, NVIDIA has just released the platform to over 40 million developers and graphic designers. This will surely power the metaverse trend forward. And it will make NVIDIA’s semiconductors critical to many metaverse projects.
So I see this as another great strategic move by NVIDIA. The company has a knack for being way ahead of the curve like this. That’s why we are up 760% on the company in The Near Future Report right now.
And we should expect to see some major metaverse developments happen next year as a result of this move. We’ll talk more about this in my upcoming 2022 prediction series.
Beeple’s back with a “digi-phizzy” NFT…
Beeple just came out with his latest piece of art. It’s called Human One. And this time, the work pairs a non-fungible token (NFT) with a physical product. That’s called a “digi-phizzy” NFT.
Regular readers know that I’ve been covering the NFT space extensively here in The Bleeding Edge. For newer readers, NFTs are digital assets that are cryptographically secured and authenticated on a blockchain. They are essentially digital collectibles.
And Beeple is a famous artist that sold an NFT for $69 million earlier this year. That was the third-highest sale for any artist alive today.
What’s interesting is that Beeple’s $69 million work was simply a collage consisting of 5,000 individual digital art pieces. Not surprisingly, this piece of digital art was titled Everydays: The First 5000 Days.
Beeple’s Human One represents the next generation of digital art. It incorporates both a physical piece of art and a digital one. And it just sold for $29 million. Not too shabby for a follow-on sale. Here it is:
Beeple’s Human One
As we can see, Human One depicts what looks like an astronaut walking through different landscapes. This is a physical installation with three-dimensional (3D) art inside the box. And it comes with its own dynamic NFT.
The NFT is linked to the physical product. It verifies that the owner owns the digital art as both an NFT and the physical installation. That’s what makes it unique.
I think we will see this approach proliferate as we move into 2022. It’s just too compelling not to catch on, as it pairs the “tangible” nature of physical art with NFT technology.
And we can see that the NFT space is still developing… and creating opportunities for savvy investors. If any readers want to introduce themselves to this trend, please go right here to learn more.
Editor, The Bleeding Edge
Like what you’re reading? Send your thoughts to [email protected].