Dear Reader,
The numbers are in… and they weren’t what industry experts expected.
Online sales figures for Black Friday and Cyber Monday rocketed to record levels over the long holiday weekend in the U.S. These are closely watched figures, as they’re believed to be an indication of the strength of the U.S. economy and consumer sentiment.
Industry experts broadly believed that sales would be lackluster this year due to the weak economic environment and repressive inflation. So what’s going on?
Black Friday sales came in at $9.12 billion, a 2.3% increase over last year. And Cyber Monday brought in an impressive $11.3 billion in sales – 5.8% above last year’s figures.
These incredible numbers come at a time of economic decline, levels of inflation not seen in decades, and an environment where disposable income is shrinking quickly due to the continued rise in interest rates.
Many have already opined that the record sales numbers are due to price inflation, or that the economy is somehow stronger than what was thought. While inflation certainly affected the numbers, that view is too simplistic and missed the bigger picture.
The reality is that retail inventories are at all-time highs. To state the obvious, that means the supply for many goods is in excess of demand.
Retailers’ Inventories
Source: Federal Reserve Economic Data
As you can see in the chart above, retailers have an excess of goods that they need to move off their books. This led to a wave of deep discounting of products this year.
Retailers know that consumers have less disposable income and that rates will continue to rise. It’s a race right now to reduce inventories as quickly as possible before things get worse.
How consumers paid for the record level of purchases is also important for us to understand.
Buy now, pay later (BNPL) purchases were up 78% compared to the previous week. BNPL purchases, if you haven’t used them, allow buyers to break up their purchase in a series of payments.
Retailers love BNPL. For the small cut they give up to BNPL players like Affirm, Klarna, or Afterpay, they can increase their average order value by as much as 85%.
That’s the dark side of BNPL: It incentivizes us to purchase more of what we probably don’t need. And it results in increasing consumer debt levels. Not good.
So it’s no surprise that U.S. credit card debt has also reached record levels this month – and is now approaching an unfathomable $1 trillion. This isn’t healthy, nor is it a sign of a strong economy.
The increased usage of credit card debt and BNPL services is simply a way to offset the loss in disposable income due to rising mortgage costs, rising auto loan payments and, of course, the dramatically increased prices of food and energy.
U.S. consumers have simply pulled in their holiday purchasing this year, and largely financed it with record levels of debt. All while knowing there will be continued rate hikes from the Federal Reserve, adjustable mortgage payments, auto loan payments, and even credit card payments that will increase accordingly.
Sadly, the Federal Reserve will interpret these numbers as an indication that it needs to continue to hike rates to destroy demand and get inflation under control. It’s short-sighted and myopic, but that’s what “they” will do.
I wish I had better news, but the Fed will continue to hike interest rates until things break. We should expect rough economic conditions and volatile markets for the first half of next year as a result.
“Digital twin” is a term that’s become popular this year.
The term refers to a digital representation of something. Maybe it’s a city. Maybe it’s a transportation network. Or maybe it’s something even larger.
“Digital twinning” is the act of accurately modeling a complex system with all of its inputs and then employing artificial intelligence (AI) and supercomputing technology to better understand it, and ultimately predict potential outcomes.
The idea is that we can ingest data and run models to simulate scenarios within the digital twin. Then we can make real-world predictions based on those outcomes.
A great example of this concept was employed last year.
That’s when Lockheed Martin partnered with the U.S. Department of Agriculture and the Colorado Division of Fire Prevention to create a digital twin of areas in Colorado that are prone to wildfires.
The inputs were things like topography, air temperature, humidity, and wind. A digital twin like this allows teams to monitor and predict in real time what will happen with a wildfire.
Well, Lockheed Martin is using that experience to tackle an even bigger challenge.
Lockheed Martin Space was just awarded a contract from the National Oceanic and Atmospheric Administration (NOAA). It’s teaming up with graphics processor unit (GPU) and AI giant Nvidia to create a digital twin of the entire Earth.
The ultimate goal? To predict the weather.
To do this, the project will ingest an incredible amount of data. This includes data around the world’s sea surface temperature, land temperatures, all kinds of atmospheric data, ocean current patterns, and all the available satellite imagery of clouds and weather patterns.
The Earth’s digital twin will process all this data in real time. Then the AI will be tasked with two things.
First, it will predict the weather in any given area. Second, it will anticipate major weather events well in advance. Hurricanes… typhoons… major shifts in weather patterns… the AI will be able to “see” all of it in advance.
This is something that just wasn’t possible a few years ago. Comparatively rudimentary attempts have been made. It’s only thanks to advances in AI and supercomputing that we can now create this kind of digital twin of the entire Earth, and more accurately predict the entire world’s weather.
This is an exciting project that’s expected to go live at NOAA by September of next year. I expect that the world will start to benefit from the results by the fourth quarter of 2023.
November is always a huge month for the auto industry. And that’s because of the L.A. Auto Show. This is where the industry’s biggest announcements tend to happen.
Last year, Fisker revealed its electric vehicle (EV) prototype, the “Ocean.” This year, a major deal was announced between Google’s autonomous driving unit Waymo and China-based Zeekr.
Zeekr is a name most of us probably won’t recognize. It’s a subsidiary of Geely, which is one of the fastest-growing car manufacturers in the world, and the seventh largest in China.
Many of us might actually be driving one of its cars. Back in 2010, Geely acquired Swedish automaker Volvo. Most probably don’t know that such a famous Swedish brand is actually owned by a Chinese company.
Zeekr is going to license Waymo’s self-driving technology for its own vehicles. This will allow Zeekr to launch an autonomous ride-hailing service in China.
Here’s a look at the vehicle:
Waymo and Zeekr’s Mobility Minivan
Source: Tech Crunch
As we can see, Zeekr’s vehicle looks a lot like a futuristic minivan. We can see that it’s loaded with sensors, the same kind that we see on a Waymo vehicle. The sensors provide the critical inputs to the self-driving AI.
This is Google’s master plan coming to fruition.
I’ve long maintained that Waymo has always been gunning to get its software into cars. We can think of its technology as the car’s operating system. It has no intention of being in the car business to all; its interest is in the software.
In my eyes, Waymo is a proof of concept to demonstrate Google’s automotive software capabilities.
It’s the same strategy Google employed with smartphones. It created the Google Pixel phone simply to showcase the Android operating system. Then it licensed out that operating system to other companies making their own phones.
It was all about propagating the software far and wide… and then collecting enormous amounts of consumer data from its use. That’s Google’s core business model. It uses this data to create a profile of us, and then it sells access to this data to advertisers.
So it begins.
This is the very first licensing deal for Waymo. It’s something that I predicted would happen before the end of this year. And we can expect that at least a couple more major deals will be announced in 2023.
The Zeekr vehicle is a great example of something designed to be fully autonomous right from the start. Here’s the interior:
Autonomous Minivan Interior
Source: Tech Crunch
Notice how there’s no steering wheel and no pedals? There’s just a tablet in the front console… and that’s it. The vehicle was designed to maximize space and comfort for passengers.
This is a view of the future of transportation. And it certainly highlights Google’s long-term strategy with Waymo.
We’ll wrap up today with another huge development in artificial intelligence.
This time it’s from Nvidia, which just released a new form of AI called Deep Learning Super Sampling (DLSS).
It sounds quite technical, but the concept is simple.
One of the biggest challenges in applications that are graphics intensive is that high-performance computing systems are required to run them well. The gaming industry is a perfect example.
The mass market tends to use gaming consoles like the PlayStation or Xbox for gaming. But video rendering can be slow, and even jerky at times on certain games.
That’s why serious gamers spend thousands – and sometimes even tens of thousands – of dollars on high-performance PCs and add-on graphics cards from Nvidia. It’s been the only way to improve performance… until now.
Nvidia’s DLSS uses software – or specialized AI – to speed up computer graphics by up to 530%. Here’s how it works…
The AI is able to predict individual pixels of graphics for an entire frame. It learns what to expect and renders those images in real time. Nvidia has demonstrated that its AI can render as much as seven out of every eight pixels.
This allows Nvidia’s technology to increase the actual frame rate of any form of dynamic computer graphics, resulting in sharper images and even better detail. Higher frame rates visually look better and are more immersive.
Here’s a look at the AI in action for image rendering:
Nvidia’s DLSS Artificial Intelligence
Source: spectrum.ieee.org
Here the AI was given eight different objects to render into computer graphics. And as we can see, it was able to create these objects from scratch in just two seconds.
Now, two months ago this process would have taken several minutes. Two years ago, it would have taken several hours.
That’s how dramatic this jump in performance is. We’ve gone from hours to minutes to seconds.
This is exciting because not only will the technology improve the performance of motion-based computer graphics – whether it be for games or enterprise applications – but it’s already improving the performance of image rendering, as we see above.
Better yet, the DLSS software also reduces the amount of computational power needed to achieve a higher frame rate. Said another way, we don’t have to throw out more and more computational power to achieve better results. The performance enhancements are being driven by AI.
So this is a breakthrough that will be absolutely transformational for producers, content creators, and game developers. And it’s another great example of how Nvidia has become an AI powerhouse over the last five years.
Regards,
Jeff Brown
Editor, The Bleeding Edge
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
