Week 30: Buybacks Over Dividends, AlphaFold Breakthrough and The First 3D Printed Bridge

Briefing: This week, three topics caught our attention. Equity yields are higher than reported. The S&P 500 dividend yield is approximately 1.25%, but when you include buybacks the yield is closer to 2.25%. Investors may not appreciate how the buyback yield is positively impacting them from both a tax and income perspective. AI is contributing to real scientific breakthroughs. AlphaFold is AI’s first tangible contribution to biology and marks a significant leap forward for drug development. Artificial intelligence skeptics say that AI does not produce revenue for any company, however, the speed of innovation in AI is accelerating rapidly. 3D printing is ready for prime time. The first 3D printed bridge was unveiled in Amsterdam last week, giving us a peek at the creativity made available through this new construction medium. Let’s take a closer look:

Buybacks over dividends

Dividend yields have gradually decreased over time to the point where they are often even considered out of fashion. The main dividend payers tend to be the older legacy companies, like Iron Mountain, AT&T, Campbell’s Soup, Wallgreens, etc. This has frustrated many investors who seek dividend yields as a way to replicate income. What these investors may not appreciate is the shift towards companies spending on buybacks instead of dividends.

Buybacks decrease the overall share count of the company, which increases the value of the shares that are leftover, including the ones that you own. Berkshire Hathaway does not pay a dividend, but they have been aggressively buying back their own shares. Investors earned a 4% buyback yield from Berkshire this year. That 4% yield is not taxable.

This practice is more common than you might think. Apple pays a .6% dividend but bought back 3% of their shares, bringing their total yield closer to 3.6%. If you look at the S&P 500, the dividend yield is roughly 1.2% but the total yield including buybacks is closer to 2.25%. My guess is that few investors who are looking for yield fully appreciate how high the S&P 500 yield actually is. The buyback yield is higher than the dividend yield for nearly 1/3rd of S&P 500 companies.

This has led to an increase in portfolio lending. Why? Well, instead of selling the accrued buyback yield, investors have turned towards borrowing against their shares at low rates, 2% or lower. That loan is used as “income”. If you would like to know more about how this works, please reach out.

AlphaFold – Deep Learning Success

Emergence is a powerful concept within complex systems theory that highlights how simple entities working together can create novel and unprecedented outcomes. AI is a perfect example of an emergent phenomenon. Google’s AI division, called DeepMind, is an emergent result of the world’s top computer scientists collaborating on top of the world’s leading computer architectures.

DeepMind recently presented the results of their AlphaFold project in Nature. AlphaFold is able to accurately predict 3D protein structures solely based on its amino acid sequence. It “learned” to do this via training on the existing relationships between roughly 100,000 known amino acid sequences and their resulting structures.

Kristof Szalay, CTO of Turbine.AI commented, “Could you predict how an airplane flies only based on an inventory of its parts? This – with proteins – is the essence of the protein folding challenge.”

Skeptics of AI say that no companies today rely on AI to generate their revenues and that it’s more of a marketing gimmick. It may be true that little to no revenue today is tied directly to work done by AI, but that is quickly changing. Prior to AlphaFold, DeepMind’s greatest successes were creating AlphaGo Zero, AlphaStar and Agent57 which could definitively beat the world’s best Go, Chess, Shogi, and StarCraft players as well as all 57 original Atari games.

This breakthrough is based more on compute power than anything. AlphaFold uses the same computing power that Google uses to power the internet. It was able to essentially brute force the AlphaFold calculation and claims to now be able to “produce every known structure of every protein known to science.” This translates directly into additional living years for patients through rapidly speeding up new drug discovery. The protein folding challenge is basically considered solved. You can expect this tool to spring up in more sectors and domains to solve other intractable problems in the near future. The timeline from start to finish for AlphaFold was just 4 years. The pace of innovation is accelerating as each new tool opens a new door to previously inaccessible opportunities.

3D Printed Bridge

The first 3D printed bridge was unveiled in Amsterdam this month. The bridge is outfitted with sensors that will monitor corrosion, shifts in load-bearing zones, environmental conditions and pedestrian use. The 3D printing is done by a robotic arm that essentially “draws” the metal structure with a welding gun. This arm was previously used to autonomously spot weld vehicles on production lines. The software that controls the arm was created by AutoDesk, a leader in CAD software used in 3D printing and modeling (CAD stands for Computer Aided Design). This type of construction has the potential to significantly reduce waste and improve weight efficiency within the construction industry. Not to mention, it can lead to radical new designs that would have been previously impossible to produce.