• Bezos’ Blue Origin gets to go to the Moon
  • An ominous new entrant in the eVTOL race
  • Alphabet’s (Google’s) medical breakthrough

Dear Reader,

That didn’t take long at all…

Last Thursday, I highlighted the largely anticlimactic debt deal – humorously referred to as the Fiscal Irresponsibility Act – which passed the U.S. House of Representatives by a wide margin last week.

Not to be overly critical of the political theater, but the name is well deserved. The Act isn’t designed to simply lift the U.S. government debt ceiling by some modest amount. It actually suspends the ceiling entirely. 

On Thursday, I predicted that we could expect a flood of U.S. Treasury issuance in the coming weeks. After all, the U.S. Treasury had been shuffling capital from one pocket to another in order to stay beneath the debt ceiling during the month of May. It seemed natural that the moment a deal had been agreed upon, it would start to fill up its coffers again.

And it appears my prediction was too conservative. 

On Friday, it auctioned $25 billion in Treasury bills. Today was an auction for an additional $65 billion, tomorrow will be $164 billion, and Thursday another $123 billion. In total, that’s $377 billion in Treasury auctions in the span of five business days. Wow!

But that’s not the most important number for us to be looking at. The U.S. Treasury uses the majority of the funds raised from the auctions to buy back Treasuries that have reached maturity. It’s basically how the U.S. government rolls over its debt. What’s interesting is how much money is left over after buying those mature Treasuries.

For the auctions over these five business days, it will be sitting on about $114 billion in net new cash.

To put this into perspective, if we assume an annual budget deficit of $1.2 trillion, the Treasury needs to raise on average $100 billion every month to keep up with the government’s irresponsible spending. The U.S. Treasury just raised $114 billion in the span of just five business days. I can’t imagine what comes next.

This level of issuance is unbelievable, and it’s highly relevant to the markets for a couple of reasons. Given the current attractive yields of Treasuries, massive issuance will suck capital out of both the bond and stock markets. It’s a continued flight to safety with a compelling yield.

This results in less liquidity in the markets. And the problem is further exacerbated by the current quantitative tightening which is happening at the Federal Reserve. It continues to pull out about $95 billion a month from the banking system.

The systemic risk in a situation like this is that the massive flooding of the market with U.S. Treasuries will force bond prices down (due to oversupply). And that means yields will rise further. That’s what we’re going to be watching out for. 

Higher yields and higher interest rates on debt are exactly what put the entire regional banking sector on the verge of collapse earlier this year. If the Fed hadn’t stepped in to offer up about $360 billion in support, it would have all come tumbling down.

This problem is even larger though. This systemic risk impacts the mortgage markets, auto loans, commercial real estate, and any other form of project financing tied to interest rates.

There’s only one key point for us to remember: When the yields spike, we should watch out below.

The billionaire space race continues…

Last week, we had a look at several prominent billionaires who have been racing to build empires in the aerospace industry. They include Elon Musk, Jeff Bezos, the late Paul Allen, and Richard Branson.

Recently, Musk’s SpaceX and Bezos’ Blue Origin have been competing directly with one another for NASA contracts. And the competition is heating up…

If we remember, SpaceX won a $2.9 billion contract from NASA back in 2021. Per the contract, SpaceX would build and launch NASA’s Lunar Gateway system as part of NASA’s Artemis program and then take astronauts down from the Lunar Gateway to the Moon’s surface.

For the sake of newer readers, Artemis is the program that outlines NASA’s plan for establishing a manned presence on the Moon. And the Lunar Gateway is the new space station that will orbit the Moon. It will serve as the command center for lunar operations.

That was an incredible win for SpaceX which out-maneuvered all of the legacy aerospace incumbents with the best proposal, as well as the best price. But Bezos was not at all happy about it.

In fact, Bezos petitioned NASA to reconsider. And he even lobbied Congress trying to get Blue Origin back in the mix.

In the end, Bezos’ Blue Origin ended up partnering with legacy aerospace companies like Lockheed Martin, Draper, and Boeing to design an alternative lunar landing system. Then Blue Origin suggested to NASA that its system be used as a backup to the work SpaceX is doing… and NASA obliged.

NASA just awarded Blue Origin an even larger contract (ironically) at $3.4 billion to move forward with its alternative lunar landing system. And NASA slotted Blue Origin in for the Artemis 5 mission. That will be the third manned mission to the Moon. It’s expected around 2029.

So we have a situation where SpaceX will support the first four Artemis missions. This includes the first two manned missions to the Moon (Artemis 3 & 4). Then, Blue Origin will be able to demonstrate its system for Artemis 5.

Either way, this is a big win for Blue Origin. It had been lagging far behind SpaceX and was desperate to win a major deal with NASA. And now there are two paths toward building a manned presence on the Moon. This certainly demonstrates NASA’s commitment to the mission.

That said, bringing Blue Origin in radically increases the costs of any future Blue Origin missions. The reason isn’t just the $3.4 billion contract with Blue Origin. The excessive costs come from Blue Origin’s proposal being tied to NASA’s Space Launch System (SLS) which costs roughly $4.2 billion for a single launch.

Perhaps the best outcome is to have the Blue Origin/SLS combination remain as a backup system. Assuming that SpaceX delivers and succeeds with Artemis 3 & 4 at a fraction of the cost, NASA can save costs and stick with SpaceX.

In some ways, this reminds me of the early days of the air travel industry. Early in the 20th century, air travel was typically only for military and industrial purposes. It wasn’t until the 1930s that air travel really became possible for leisure travel. Even then, it was expensive… and uncomfortable.

But as the industry matured, air travel became more common for everyday people. That’s especially true in the 1970s when we saw an explosion of competition and new entrants into the space. Today, traveling by plane is commonplace due to how affordable trips have become.

It’s developments like these that make me believe we will see a similar evolution in the space industry. Right now, this new generation of space transportation is mostly for government and industry. But we’re already seeing the early signs of passenger space travel.

And one day in the not-so-distant future, some of us may be planning whether to go on a safari or take a vacation to the Moon.

Archer set to compete at the price of an Uber ride…

We’ve been tracking electric vertical takeoff and landing (eVTOL) company Archer Aviation for two years now. The company made a big splash in February 2021 when United Airlines placed a $1 billion order for Archer’s eVTOLs.

Well, Archer Aviation just had its quarterly earnings release… and something interesting caught my attention. Management announced that they hope to get their Midnight eVTOL certified by 2025. We’re no more than two years out.

That means we’re getting very close to air taxis being a reality. And I have to say, the Midnight certainly looks sharp. Here it is in action:

I have to admit, when I first saw the Midnight, I thought it had a striking resemblance to the AI-powered drones from the movie Terminator. It even has the pointed “nose” to resemble a predator, like the beak of a hawk.

Source: 20th Century Studios

Fortunately, the Midnight isn’t designed for military applications. And as we can see, the Midnight can effortlessly take off vertically from landing pads in an urban environment.

The craft can carry five people. That’s a pilot plus four passengers. It has a range of about 100 miles. And it can reach up to 150 miles per hour.

And get this – Archer Aviation believes that its eVTOL flights will be cost-competitive with ride-sharing programs in urban settings. That is to say, a 30-mile eVTOL flight will cost about the same as a 30-mile Uber or Lyft ride into the city.

The difference is, eVTOL flights will get passengers to where they are going much faster. That’s because they can fly from Point A to Point B in a straight line. And they won’t have to contend with traffic.

So if we think about commutes from the suburbs to a major city – 30-mile drives can often take over an hour due to traffic. For comparison, an eVTOL flight would take just a few minutes.

For this reason, I believe we’ll see eVTOLs become a popular mode of transportation in urban areas very quickly. I believe that there will be incredible demand for this kind of service. And it’s all going to happen within the next two or three years starting in the larger metropolitan markets.

Google’s large language model leaps ahead…

Google just released the second version of its Med PaLM 2 generative artificial intelligence (AI). And it’s made a huge leap over the first version. This has big implications in the medical field…

As a reminder, PaLM stands for “Pathways Language Model.” It’s a large language model similar to OpenAI’s GPT-4.

And as the name implies, Med PaLM is the same kind of generative AI, and it’s trained on medical knowledge. The idea is that it should be able to answer medical questions and ultimately diagnose patients given enough information. And it’s quite good…

Source: Google

This graph charts how various generative AIs performed on the MedQA dataset. This is a series of questions written in the style of the U.S. Medical Licensing Exam (USMLE).

As we can see, the first version of Google’s Med PaLM model answered these questions with 67.2% accuracy. At the time, that was the best performance in the industry to date. That was better than GPT 3.5’s performance. It performed with 60.2% accuracy.

And look at Med PaLM 2. The new model answered medical questions with 86.5% accuracy. That’s remarkable.

This is another fantastic example of how quickly generative AI is advancing. And to think – this big jump came in a matter of just five or six months. That’s remarkable.

The use of generative AI for a field like this is so practical. In a healthcare setting, it can essentially augment or replace human doctors if none are available, even to the extent of diagnosing patients’ illnesses.

And it’s easy to envision how Google will take this to the next level. Imagine being able to feed the AI model individual patient data – various tests, blood pressure readings, imaging results… whatever it may be.

With this data, the AI could offer custom-tailored diagnoses specific to each patient. And it could even make potential treatment recommendations as well.

A physician wouldn’t need to be present for this. Instead, a physician assistant (PA) or a nurse could collaborate with the AI directly. This would be an incredible resource for the industry, especially in underserved markets.

That’s the direction this technology is heading. Google is set to collaborate with various medical institutions and organizations to beta test this model.

What’s more, Google will almost certainly release the next version – Med PaLM 3 – by year’s end. I’m predicting this new version will be capable of processing specific patient data. This is part of a hot trend in AI right now called multimodal AI. And that’s when we’ll be off to the races…

So this innovation is going to be an invaluable resource for the medical community. And it stands to democratize access to world-class healthcare. Suddenly, patients in developing countries will have access to the best and most current medical knowledge on the planet.