Editor’s Note: This month our Spotlight is focused on George Gilder, the man Porter calls “the prophet” and says was instrumental in shaping much of his economic thinking and analysis.
Last week we introduced you to Part I of George Gilder’s New American Century thesis (click here to re-read) and why he believes we’re about to enter a new era of American prosperity.
Today, we’re sharing Part II, which explores the invention that Gilder says could be the next world-changing technology, and the companies at the frontier of this new revolution.
While the team at Porter & Co. remains bearish and expects a severe pullback in the short-term, there is no doubt that long-term the future is bright.
Although we’re not in the business of predicting the future – and prefer to focus on undervalued “forever” companies with the potential to compound at 15% to 20% annually – we also recognize that intelligent speculation can deliver unparalleled returns (see Porter’s 1990s recommendations of Amazon, Adobe, Texas Instruments, and others).
George Gilder is one of the best in the world at finding these types of high-risk, high-reward moonshots. If you’d like his full portfolio for this New American Century, you should read this letter he wrote.
The nation which enjoys the most freedom must necessarily be in proportion to its numbers the most powerful nation.
— John Quincy Adams
Last week , I laid out why I believe we are entering a new era of prosperity and innovation not just in technology, but across America. I also shared some of the reasons why I am so optimistic about our future.
I believe this new era will be The New American Century.
And I say ignore the naysayers who I call…
THE ARMY OF DESPAIR
The pessimists themselves.
You can tell a pessimist easily. He’s the one who calls the cops. The pessimists don’t believe we can compete with China, so they forbid American companies from selling to China, forcing U.S. companies off the learning curve and effectively defunding their R&D. Then, addicted to the substitution of power for knowledge, they tax Americans to subsidize government-directed R&D programs a decade behind the times.
The pessimists talk of war but fund a military strategy dating to WWII.
They don’t believe American ingenuity can solve our energy challenges, so they subsidize “renewables” that don’t renew, currently at a cost of some $2 trillion a year.
Everything the pessimists advocate comes down to managing for failure. In just the past two years, they have sacrificed most of America’s technology lead in semiconductors by imposing sanctions on China that, unsurprisingly, have achieved the opposite of their intent.
THE FATAL LUST FOR CERTAINTY
The pessimists yearn for certainty and make only misery. The pessimist is the man who wants to “make sure,” and since he knows only the past, imprisons us there. Uniting the Communists, the Nazis, the Greens, and socialists of every stripe is despair and denial of the providential God who made a bountiful universe and gave us the minds to comprehend and cultivate it.
We do not believe they will prevail or that America will decline. John Quincy Adams was right. It is liberty that makes a nation powerful and a people great. Despite enlarging encroachments on that liberty, our traditions of freedom are deep and strong. That tradition, that American habit of mind, even now gives us an inestimable advantage in maintaining both our freedom and our power as compared to rivals habituated by centuries, even millennia of autarky.
This November, I will be gathering a league of America’s greatest and most inspiring optimists: entrepreneurs, inventors, scientists at the annual COSM conference of the Discovery Institute. This year’s theme is “The New American Century.” Come, have your faith renewed, and get the inside story on some of the world’s greatest new companies.
THE SPATIAL WEB GETS REAL AND NVIDIA IS THERE
What is the Spatial Web anyway? More than one reader has asked this recently. “We get,” they say, “that it is big and bold and important and the successor to the Internet and you guys think it will change the world, but you haven’t really explained what it is.”
Partly this is because descriptions of virtual phenomena tend toward abstraction and partly because the Spatial Web is still emergent; examples have been few and fragmentary.
That’s changing, quickly. Some of the most important companies in the world are working in the Spatial Web right now, designing virtual factories, engines, or microchips before laying out the capital for construction.
Taking the next step, companies like Foxconn are going beyond mere modeling (however amazingly true to life) and using the virtual factories to train the robots that will work in them.
Foxconn is the largest contract electronics manufacturer in the world. The Taiwanese company does not make chips, but it builds products that use chips, including two out of three iPhones. It has something like 170 factories globally.
Our interest is in the company’s 171st factory, which is located—nowhere or, if you like, everywhere. The 171st factory is the “digital twin” of Foxconn’s new factory in Guadalajara, Mexico. Digital twins are an essential element of the Spatial Web.
Using advanced versions of the same technologies that make modern video games so realistic, engineers build, for instance, virtual versions of a new factory. They can test it out under various scenarios from an earthquake to a blackout to a sudden upsurge in orders. They can see how the factory performs at levels of detail impractical with the physical version of the building.
Designing a modern, automated assembly line, engineers must correctly place dozens of robot arms—strong, big, and bulky weighing many hundreds of pounds. Digital twins allow the engineers to test, modify, and retest their arrangements in ways that are impossible using the physical machines themselves. They can simulate modifications of the design and assess whether these are improvements. They can get accurate estimates of needs for material and labor to build the system and discover how to reduce both.
A “physics-based” digital twin knows the reality of the factory, or the new model car (Volvo) or the new aircraft engine (Rolls Royce), or the new drilling platform (Shell) or even huge infrastructure projects (London’s Crossrail) from “inside out” at a level of detail unavailable in the physical world. No team of inspectors of a physical facility has ever had as complete a view of a large, bulky, complex system as the digital twin can provide.
Digital twins are more than design tools. They can tell the humans what is going on in the physical system in real time. At the physical version of the system, engineers install (with guidance from the digital twin itself ) arrays of sensors—of heat, pressure, light, air quality and more. In a complex system such as an automated factory, the sensors will number into the thousands. The data gathered are so voluminous that no human being could effectively track them. An artificial intelligence machine does that and tells the humans in real time when something is going wrong.
Rolls Royce builds a digital twin of every engine it makes. Sensors in the physical engine track every relevant aspect of performance, the AI updates the digital version in real time. Instead of waiting for a failure, the twin uses sensor data to catch problems in advance and enable “predictive maintenance.”
DIGITAL SQUARED
Now Foxconn is taking all this a step further. It is using a digital twin of its factories to train other digital machines—AI robots—how to run the factory. Like a human pilot in a flight simulator, the robot arms practice, e.g., how to see, grasp, and position components. The factory’s fleet of AMRs (autonomous mobile robots) will practice navigating the factory floor inside the digital twin.
At the root of this entire effort is Nvidia. Its Graphical Processing Units (GPUs), invented for video games, are the essential tool for making digital twins. Along with Siemens, which provides its Xcelerator hardware and software package to be integrated into Nvidia systems, Nvidia provides most of the critical supporting technology as well.
- NVIDIA Isaac™ Manipulator provides NVIDIA-accelerated libraries and AI models that enable AI-enabled robot arms that can perceive, understand, and interact with their environments for machine tending, inspection, assembly, and even bin picking.
- NVIDIA Isaac™ Perceptor provides advanced perception capabilities enabling autonomous mobile robots (AMRs) to “see,” “feel” and operate in unstructured environments such as factories.
- NVIDIA® cuOpt™ A big advantage of digital twins is the ability to optimize moving stuff around, e.g., a factory floor, by picking the best routes or re-routes, balancing loads and foreseeing all the stuff that could go wrong when moving big, heavy, expensive objects in a crowded and potentially perilous space. Traffic jams are costly and dangerous. NVIDIA cuOpt™ minimizes them. As Nvidia likes to point out, “with 23 world-record benchmarks, cuOpt owns all of the world records on the largest routing benchmarks from the past three years.”
- NVIDIA Isaac Sim™ is what enables developers to train AI-based robots within the digital twin autonomous machines in a physically based virtual environment.
- Metropolis Microservices™ combines multiple AI-driven capabilities to design and manage smart buildings—or cities.
At the base of all these technologies sits the NVIDIA Omniverse™, a comprehensive development platform for digital twins. It simplifies the use of technologies such as Universal Scene Description (originally developed at Pixar to make their animations more realistic and efficient) and RTX (real time rendering) to update the digital twin against physical reality.
Which brings us around to the original question:
Q: What is the Spatial Web?
A: It is what you get when the internet delivers not merely words or even moving pictures but faithful, dynamic, real time, digital representations of the real world that provide actionable and accurate information about a material object or system. This information is often far more complete than what could be gleaned from direct examination of the physical object. It allows us to represent the physical system virtually and then create that system in reality.
Q: Okay, but you just explained that’s what digital twins do. How is the Spatial Web different?
A: The Spatial Web is what happens when all these digital representations are networked to allow them to interact with each other. Now if I want to order my flying Thunderbird, my AI can consult with the digital twins of foundries, warehouses, design shops, railroads, steamships, and truckers to plan it all, check in with my bank and my government regulators to approve, and then push the “go” button.
We are not quite there yet but progress is faster than we expected. What seems likely to happen is that all this will become a reality so seamlessly that most people will barely notice it until it defines the world.
CUI BONO (WHO BENEFITS?)
There is hardly a company we cover that will not benefit from this next step into a world impregnated with machine intelligence. And machine intelligence itself will benefit—because even now microchip designers are using digital twins of those devices to simulate and select the best combinations of features. Assembling a system from various chiplets, for instance, requires understanding not only the logical but physical relationships across the merged devices and the implications of the technology used to join them. Digital twins can do that work more effectively and efficiently than an engineer and his equations.
These developments, however, do have special relevance for two companies we cover, Nvidia and Verses AI. The latter is the subject of this week’s Moonshots.
As for Nvidia, the debate rages on: Can it possibly grow fast enough to justify even its current $3 trillion valuation at an astonishing 40X sales and 70X earnings? How could it possibly keep that up?
Viewed as a maker of GPUs, the essential chip for AI, then the likely answer is that it can not. Competition is on the way. Data centers are hunting desperately for ways to economize at NVDA’s expense. We expect the current unearthly growth rates in NVDA’s GPU sales to continue for maybe a couple more years then slow. When they slow, expect NVDA to go on sale, big time.
Nvidia, however, is changing. It is adapting to the new world that its own devices have created. Nvidia does not merely make the widgets, it shows the way with its software libraries, courses, and consulting services. It is becoming the essential intellectual partner for enterprises from heavy industry to Hollywood to health care. If it continues to hold that privileged position, growth could even accelerate. Revenues may continue to take the form of orders for GPUs even as NVDA becomes one of the globe’s most important consulting firms, showing companies how to engage in the world of the spatial web.
IT’S HAPPENED BEFORE
Amazon was a mature though still growing firm in 2017. Then in the next three years its revenues more than doubled. By 2021, its earnings were up by 10X. The difference was Amazon Web Services (AWS), which boosted margins even more dramatically than revenues.
AWS made Amazon a different company. But here is the coolest thing. The change was enabled by Amazon’s existing, low profit business. Amazon needed the hardware— data centers and all— that became AWS just to run its own business. That asset then created a new demand that made AMZN a crucial partner to global enterprises. Mr. Say strikes again.
By creating the GPU, Nvidia has changed the world into which it sells GPUs. We are not rushing out to buy more of the stock. We continue to keep it on a hold rating with the requisite 1% allocation. But if (when) it goes on sale, we will be there.
UPDATED NVDA VALUATION MODEL
The stock has declined considerably since we downgraded it to a Hold but the valuation model is still consistent with that rating.
Until Then!
George Gilder
Editor, Gilder’s Technology Report & Gilder’s Moonshots