• Democratizing education – personalized tutors for every student on Earth…
  • Microfactories spur decentralized manufacturing…
  • China takes the lead with hyperloop technology…

Dear Reader,

He literally threw in the kitchen sink…

Whatever it takes to get the deal done.

Elon Musk Visits Twitter HQ

Source: Twitter

Yesterday, Elon Musk stopped by his new Twitter headquarters in San Francisco, California. He carried a kitchen sink with him and posted an accompanying tweet saying: “Let that sink in!”

It’s a hilarious end to the spectacle that became Musk’s acquisition offer for Twitter. 

The back-and-forth negotiations and refusal by Twitter revealed the problems of bias, manipulation, suppression of free speech, and fake bots at Twitter are far worse than most expected. Hopefully, we’ll find out just how bad in the months ahead.

Yes, the deal will be done. Oddly, there are still skeptics even though the court-issued deadline on the deal is tomorrow, October 28.

Most of the skeptics are coming from the heavily biased media. I’m not one of them. The terms of the deal are set, the contracts are signed, and all of the cash is already in an escrow account. Game over.

Sending the money is just a formality, which is why Musk brought in his kitchen sink on Wednesday.

What most don’t know is that Musk tweeted back in June: “I’m dressing as a sink on Halloween, as they will have no choice but to let me in.”

That was a remarkably prescient comment considering the October 28 deadline hadn’t been set at that time, and Halloween is on Monday. 

Let that sink in…

Now, the real work is ahead. Those who disagree with the first amendment of the U.S. Constitution aren’t going to like what comes next. Those who favor political “sciency stuff” over real scientific research are also going to hate the deal.

For the rest of us, it will be a wanted restoration of certain individual rights and freedoms.

The panic that’s set in amongst the media and those at Twitter who believe they’re violating individual rights is palpable. This won’t end well for them.

Musk has already been clear that he believes Twitter is bloated and way overstaffed, and he’s right. He’s even suggested that he’ll lay off as much as 75% of the workforce once he takes control.

I’ve been through more restructurings than I prefer to recall. They’re always difficult. And they’re always necessary.

Twitter’s business was flailing. Its mismanagement and loss of focus presents an incredible opportunity. And if it continued down its current path, it would have died a slow and painful death. 

Come Monday, Twitter has a new lease on life. We should expect the radical changes and improvements to happen quickly. After all, Musk has been preparing for this all year.

And I know what he’ll be doing… He’ll be putting that sink to good use by cleaning up the dirty dishes.

An AI designed to learn like humans…

One of the most challenging problems to solve for an artificial intelligence (AI) is teaching it how to learn like a human.

It involves both the use of neural networks and specialized training environments that facilitate learning. When the industry masters this, the potential for AI is nearly unlimited.

An AI company focused on building this same kind of technology, called Generally Intelligent, just came out of stealth with a $20 million raise. It’s always exciting when these high-tech companies pull back the veil and give us a view of what they’re working on. We’ll talk about that in just a minute.

But first, we have to note that Generally Intelligent has some very strong backing. OpenAI’s former GPT-3 lead and the former head of OpenAI’s robotics division are both behind this company.

And they put in place an interesting financing structure.

In addition to the first $20 million raise, the team has secured commitments for an additional $100 million. It will be released in tranches as the company hits specific milestones.

This is very unique in the venture capital (VC) space. It’s like packaging a Series A and a Series B round together in a tranche structure.

What makes Generally Intelligent so interesting is its approach to AI. The team is focused on general-purpose AI. Hence the name.

The company’s approach is to design AIs that can learn the same way humans learn. It does this by immersing its AIs in very complex, three-dimensional environments.

The nuance here is that general-purpose AIs can learn and work on any task. That makes them useful for many potential applications.

And Generally Intelligent announced that it will deploy AIs for two standard applications: scientific discovery and new material design. These have been a common focus in the industry.

But what I’m more excited about are the company’s two other areas of focus: personal digital assistants and AI tutors.

I’ve been predicting the rise of personal digital assistants for the last few years.

We aren’t talking about basic assistants like Alexa and Siri here. The next generation of digital assistants will be able to do far more complex tasks.

They will book our appointments… make our dinner reservations… order our groceries… Maybe they’ll even do our taxes.

They’ll be accessible to us through our smart speakers at home, our smartphones, our smart watches, even our augmented reality eyewear.

In short, these digital assistants will handle our menial tasks for us. And that should free up at least an hour of time every day for most consumers.

And if we think about AI tutors – they could be transformational.

Imagine a tutor that can learn the absolute best way to teach any subject to an individual student. Imagine a tutor with a completely optimized one-on-one tutoring style based on the student’s ideal way to learn.

This just isn’t possible within our educational system today. But with technology, we could give every student in the world their own personalized AI tutor. That would transform education overnight.

So this is a radical idea that I’m very excited about. In fact, I’m quite passionate about it, as it will democratize access to the very best education in the world to pretty much any child on the planet regardless of their background.

Looking forward, Generally Intelligent will release its large 3D learning environment at the NeurIPS 2022 conference late in November. I expect we’ll learn even more about the company’s approach and its progress after that.

This is a company that we’ll hear a lot more of in 2023.

Recalibrating to decentralized manufacturing…

Radical changes are taking place in manufacturing that look completely different than what we’ve become used to over the last few decades.

A European company called Arrival just provided us with a great case study. This is a company that’s a relatively new entrant in the electric vehicle (EV) space.

Arrival went public last year with a very specific strategy. The company’s first focus is on building electric delivery vans. Then it wants to expand into electric buses for public transport. And eventually, Arrival plans to produce passenger EVs.

Up to this point, Arrival had its manufacturing center in the United Kingdom. But the company just announced that it’s moving production from the U.K. to the United States.

There are two primary reasons for this.

First, the U.S. is a radically larger market for delivery vans. Companies like UPS and FedEx are actively looking for electric delivery vans right now. So it makes sense that Arrival would want to focus on this market.

The second reason comes down to incentives. The $737 billion Inflation Reduction Act that passed over the summer provides very generous credits for EV production on U.S. soil. The credits range from $7,500 to $40,000 for commercial EVs.

This serves as a form of stimulus for EV producers. It makes their products cheaper for those who want to purchase them. In theory, it results in higher sales and better margins.

So Arrival is moving to the U.S. to capitalize on these incentives. But the company isn’t employing the old model where it builds one or two giant factories.

Instead, Arrival will deploy what it calls “Microfactories.” These are small, efficient factories that take up less space, are less expensive to build, and less expensive to operate.

This table spells it out:

Arrival’s Microfactories Versus Volkswagen Factories

Source: Arrival

Here we can see that it takes 10 of Arrival’s microfactories to produce the same number of vans as one of Volkswagen’s traditional factories. But look at the efficiencies…

A single traditional factory occupies 2.2 million square meters (sqm) of space. Meanwhile, 10 microfactories take up just 200,000 sqm.

And that makes the microfactories much more capital efficient. It only costs about $440 million to build 10 of them. That’s nearly half the $950 million it costs to build a traditional factory.

The microfactories cost half as much to operate each year as well. That’s $120 million compared to $240 million.

And lastly, Arrival can get its microfactories up and running in just six months. Meanwhile, it takes 36 months – three full years – to open a traditional factory.

The difference is pretty remarkable. And there are many more benefits as well, like reduced logistics costs and import taxes just from being close to your end markets.

This approach enables the company to locate production in all its key markets. That reduces costs and makes for a far more resilient supply chain.

This is a great example of what the future of decentralized manufacturing will look like. Small, efficient factories located close to the end markets.

I’m very excited to see innovative companies adopting this kind of strategy. I think we’ll see a lot more examples like this in the months to come.

That said, please note that I do not recommend investing in Arrival right now.

I love what the company is doing with its manufacturing strategy. But it hasn’t turned the corner from a business perspective yet. We would need to see that happen before considering Arrival as an investment target.

Conceived in America, realized in China…

We’ll wrap up today with an interesting update on hyperloop technology.

If we remember, Elon Musk conceptualized this tech about a decade ago. He was so frustrated by the traffic and the public transportation system in California that he envisioned a better way. 

So, Musk designed hyperloop technology as a way to build high-speed rail transportation at a fraction of the cost of what was being envisioned by the California state government at the time. Then he detailed exactly how a hyperloop system could work in a research paper.

The concept is simple. It involves low-vacuum tubes that employ magnetic levitation technology.

The low-vacuum environment enables objects to move through the tubes with very little resistance. This reduces the amount of energy required to build and maintain speed. And that allows the magnetic levitation tech to move objects incredibly fast through the tubes.

It’s a really cool technology that could be built for a fraction of the cost of high-speed trains. And we did see a few companies take a stab at it here in the U.S., like Hyperloop One and Arrivo.

But sadly, not much progress has been made. Arrivo went out of business and Hyperloop One was acquired by Virgin Hyperloop, which has now refocused on using the technology for transporting cargo. Yet it hasn’t done much at all.

China, however, has taken this concept and run with it.

A collaboration between the North University of China and the Chinese government just demonstrated a hyperloop train moving through a 1.25 mile test track. It reached speeds of 80 miles an hour.

They’re now constructing a 60 kilometer (km) test track for further testing. That’s about 37 miles. And this will allow the hyperloop trains to run at speeds of 1,000 km per hour, or 621 miles per hour.

That’s the speed of an aircraft… on the ground. What a time saver that would be.

So, this is a remarkable development. China is proving that the tech works. It’s just a shame this isn’t being actively developed in the U.S.

After having lived in Japan for as long as I did, I’m amazed at how poor the U.S. public transportation systems are. The U.S. is at least 50 years behind a country like Japan.

Japan had bullet trains before I was born. They’re awesome to travel on and remarkably convenient and pleasant.

I only hope that one day we see Musk’s vision come to fruition here in the States. If no one steps up to make it a reality, I wouldn’t be surprised to see Musk do it himself. 

In fact, the concept is already listed as a project of one of his companies, The Boring Company. So hopefully big plans are ahead.


Jeff Brown
Editor, The Bleeding Edge