Digital twins and The Flow

"Imagine that we could predict the future. Digital Twins are virtual models that mimic the real world and let us see the future to better understand the impact of today's actions on tomorrow's world" - Jensen Huang.

Jensen Huang , co-founder and CEO of NVIDIA, is known not only for his distinctive style but also for his technological vision.

Mr. Jensen has had quite a few bright ideas throughout his lifetime - some of them happened to turn the tech market upside down:

• Back in the 90s, he went all-in with graphics processors (GPUs), which by the mid-2000s were all the rage.
• In 2000s they scaled graphics and released CUDA - programming interface that allowed utilizing GPUs for general processing workflows.
• CUDA in turn has powered crypto revolution in 2010s, that contributed to immense scale in graphics processing power which just happens to be what current generative AI revolution needed. Quite a ride, huh?
• Today NVIDIA is all about handling immense AI workloads. This helped the company become one of the most valuable tech companies in the world.

There is one more thing that Mr. Jensen is passionate about: a digital twins. In a nutshell, digital twins are a digital replicas of real world physical systems. When simulated in virtual environment they help to predict behaviors, failures and optimize products for specific conditions.

I find it well explained in the video below, where NVIDIA shows off fully simulating the warehouse:

The second Earth

While simulating warehouses is profitable it is just a tip of the iceberg. NVIDIA moved on to something much grander - simulating weather. Weather simulations hit the jackpot as both being incredibly important and historically difficult to do. Well until now. Below is a great demonstration how NVIDIA simulates not only the weather, but also environmental impacts of it.

NVIDIA calls this model "Earth 2". It is also a digital twin, but instead of a warehouse it is a model earth itself. Currently these simulations are limited to simulating weather in particular regions and geographies. For now.

Looking beyond weather and warehouses

When scaled sufficiently such technology could go much further than weather predictions or warehouse simulations. What if we could accurately simulate anything?

Sensors are everywhere. From humidity and temperature sensors on every communications antenna tower, to surveillance cameras, to your phone.

In theory, feeding a continuous stream of data to a model like Earth 2 would allow it to predict the flow of any data points that were previously measured, simulating a complex relationships between them.

Visualize that instead of the weather patterns you could visualize people flow. Commuting, socializing. Purchasing goods and services. Forming views and preferences. Spreading diseases. But instead of segregating to specific clusters this would be done to the accuracy of one person. A digital twin of YOU. Your unique Flow amongst the Flows of many others.

Proactively simulating local events at global scale.

When a weatherman says it could be raining today, you take an umbrella before leaving home. In smart cities, when simulations predict a traffic jam, the traffic light schedules are adjusted throughout the city to accommodate the car flow. These are good examples illustrating how a reactive approach helps avoid simulated scenarios.

Our world is measured with ever-increasing accuracy and frequency. This data, when processed, provides a unique opportunity to observe and visualize patterns of life itself. With a little help from cutting-edge tech, it is possible to simulate these patterns, helping to make accurate future predictions.

This data, combined with ever-increasing compute power, allows for real-time inference and opens the door for unique proactive approaches. What if instead of asking the model "will it rain?" one could ask "what's the easiest path to make it sunny?". Or, maybe even - "how to stop the spread of a pandemic?".

I believe that this is what Earth 2 and similar models are building up to solve.

Controlling The Flow

It's about harnessing the power of digital twins, vast data streams, and cutting-edge simulations to shape our world proactively. Instead of reacting to traffic jams, we could prevent them before they form. Rather than bracing for economic downturns, we could reroute the Flow towards prosperity. The potential is immense - orchestrating the very rhythm of our existence, all through the lens of predictive technology.

The convergence of digital twins, vast data streams, and compute is opening doors to a future where we can not only predict but potentially shape the Flow of our world.

From weather patterns to economic trends, from traffic flow to pandemic spread, the applications seem boundless.

Mr. Jensen isn't just anticipating this future—he's actively building the tools to make it a reality. As these technologies continue to evolve, they may well redefine our understanding of what's possible, blurring the lines between simulation and reality itself.