Dear Reader,
Welcome to our weekly mailbag edition of The Bleeding Edge. All week, you submitted your questions about the biggest trends in technology.
Today, I’ll do my best to answer them.
If you have a question you’d like answered next week, be sure you submit it right here.
Let’s begin with a question on the convergence of artificial intelligence (AI) and biotech:
Hi Jeff,
Very interesting read about DeepMind’s newest creation, AlphaFold. However, as this might now become the go-to, open-sourced AI supporting drug development for whomever employs AlphaFold, how do you see this impacting existing biotech-AI service companies? Thank you for an always fascinating service.
– Philipp V.
Hello, Philipp, and thanks for sending in your question. The latest developments regarding AlphaFold are incredibly exciting.
For new readers, Google’s artificial intelligence (AI) subsidiary DeepMind announced that its latest AlphaFold software can accurately predict the folding of a protein, based solely on its amino acid sequence, with 92.4% accuracy.
In essence, proteins are long chains of amino acids. Depending on the order of the amino acids, these proteins serve different functions. In fact, our bodies use tens of thousands of different proteins, which are responsible for every function our bodies perform.
And all proteins fold into specific, three-dimensional shapes in order to function properly. Unfolded or misfolded proteins contribute to many diseases.
Additionally, we can learn a lot from a protein’s structure. The structure tells us how it will impact living organisms. It also determines what other compounds (such as pharmaceuticals) the protein can bind to. With this information, we can design the perfect biopharma therapies for any possible ailment.
However, proteins fold in very complex ways. This makes a protein’s structure hard to analyze. Drug design has remained largely a manual, trial-and-error process for this reason.
And that is why AlphaFold holds so much potential for the biotech industry. And as I wrote recently, DeepMind just made the source code of AlphaFold2 – the second generation of this software – public. That means the entire biotech industry can begin using it immediately.
DeepMind also released a database of all the predictions AlphaFold2 has made to date – all 365,000 of them. And it plans to release an astounding 130 million protein predictions by the end of this year.
And this is great news for the biotech industry as a whole. We don’t need to worry about other companies working at the convergence of AI and biotech falling behind because of AlphaFold2’s release.
In fact, this will likely be an aid to many. Because the code is now open source, existing biotech companies will be able to use this research to inform and refine their own projects. This could even speed up their progress on therapies targeting some of the most challenging diseases out there.
I expect we’ll also see new companies form to advance new therapies leveraging these protein predictions.
I’ve never been as excited about this trend as I am now. We’re going to cure diseases – and have incredible investment opportunities – as a result of this move.
And I encourage anyone who’s interested in investing in this trend to go right here to learn about my current recommendations in this space.
Next, a reader wants to know more about iron-air batteries:
Jeff: As always, enjoyed the issue. Was fascinated by your reference to the “iron batteries.” We use nickel-iron batteries as the storage device on our solar installation. They are also known as “Edison batteries,” as they were invented by Thomas Alva Edison, and some that he built still function today.
The per cycle cost is far lower than either lead-acid or lithium-ion batteries, but the upfront cost is far higher, for utterly inexplicable reasons. As you note, iron batteries are cheap. They will last essentially forever, if properly maintained. They are very bulky. We have 2250 aH storage capacity, and they can be run down to 0% repeatedly without harming them. [That’s] contrary to lead-acid, which is damaged at 50% and lithium-iron at 30%.
Lead-acid [batteries] have a service life of maybe seven years… lithium-ion, maybe 15. Would appreciate you advising how the new “iron batteries” are different than what we have had for over 100 years.
– Walt H.
Hi, Walt – thanks for sending in your question. Battery technology has many real-world applications, so it’s always exciting when we see advances in this space.
As a recap, I recently wrote about Form Energy’s brand-new “iron-air” battery. This battery uses an innovative membrane technology to charge and discharge the battery by repeatedly rusting and unrusting small iron pellets.
Iron is abundant and cheap. Form Energy believes that its iron-air batteries will reduce the cost of long-term energy storage to $6 per kilowatt-hour.
As a comparison, lithium-ion batteries enable energy storage at $50–80 per kilowatt-hour thanks to the expensive minerals they need like nickel, cobalt, lithium, and manganese. So Form Energy’s product represents a dramatic cost reduction.
Form Energy also says its battery can be used continuously over a multiday period – potentially providing power for up to 150 hours. And it could help solve the problem with the intermittent supply of renewable energy with solar and wind farms, for example.
This means these iron-air batteries could be good replacements for many of our lithium-ion batteries, especially in industrial applications or even residential home applications like yours.
And as for your question, the way these “iron-air” batteries operate is different from Edison “nickel-iron” batteries. Edison batteries operate with a nickel-positive electrode, an iron (or sometimes cadmium) negative electrode, and a potassium hydroxide solution.
They were originally developed as a replacement for our traditional lead-acid batteries. As a lead-acid battery charges and discharges, for example, its lead plates dissolve and are reformed, but they never recover to their original condition. Sulfate crystals build up over time on the lead plates, which eventually kill the battery.
The electrodes of a nickel-iron battery, on the other hand, are insoluble in the electrolyte, so it doesn’t degrade in the same way.
And Edison batteries do offer many advantages as a result. These batteries are robust and can handle being discharged or overcharged without harm. They’re long-lasting and recyclable, as they don’t contain lead or other heavy metals. These features have made them a good choice for off-the-grid applications like railroad signals, and they’re also popular in trains and mining.
On the flip side, Edison batteries are larger and more expensive to manufacture – they can be as much as four times the price of lead-acid or lithium-ion batteries. They also have low energy density and don’t retain their charge as well as other alternatives. And their performance is limited in low temperatures. They also release hydrogen, which means ventilation is an important consideration.
These are some of the reasons they’re not used as much as the alternatives I mentioned earlier. That said, various projects have been looking to modernize the Edison battery, solve some of these problems, and bring it into more popular use. And I certainly think it has potential.
Ultimately, there is space for many innovations in battery technology that will be specific to the application for which the battery is used.
We’re going to keep a close eye on Form Energy and see how it develops its technology. After all, the company is showing great progress, but it’s not at the stage where it is commercializing the technology. At the end of the day, that’s what counts.
Let’s conclude with some questions about vision-based self-driving cars:
I note that Tesla and other car companies are going to visual inputs vs. radar and such. How well will these vehicles be able to operate in the dark (should you encounter a vehicle with no lights), in the fog, very heavy rain and if their cameras get dirty? Seems there could be issues here.
– Brad
Hello, Jeff Brown and friends,
The autonomous vehicle industry seems to be going to vision-only, but how does that fare in rain and snow, especially the snow that sticks to the car as we drive? Thank you all for your hard work.
– Jason A.
Hi, Brad and Jason. Both of you had similar questions, so I wanted to address them together. Thanks for writing in on this interesting topic.
Last month, I shared how Tesla had launched beta version 9 of its full-self-driving (FSD) software. Tesla had rebuilt its AI from scratch, basing it on neural networks.
And as you both noted, this latest version moves away from AI relying on radar and sensor data. Instead, it focuses on video input from the car’s cameras. This models the car after the human brain. When human drivers are behind the wheel, we operate exclusively based on vision too.
This FSD software provides advanced driver assistance that reduces the more burdensome parts of driving, but it’s well on its way toward a completely autonomous self-driving experience.
Some skeptics have questioned the safety of dropping radar and LIDAR (light detection and ranging) in favor of vision alone, but Tesla CEO Elon Musk reports that the vision system has improved so much, Tesla cars are better off without radar.
In fact, while radar and LIDAR may appear to be critical technology, computer vision alone can have improved resolution and less noise. Even in the dark, this technology can see with better visual acuity than the human eye.
That said, Tesla vehicles are also equipped with ultrasonic sensors that supplement the car in “seeing” any physical objects including cars, trucks, or pedestrians.
And the change seems to be supported by the fact that the Tesla Model 3 quickly regained its status as a Consumer Reports Top Pick following an independent test of the new vision-based system. And the Insurance Institute for Highway Safety evaluated the new system as well, granting it a Top Safety Pick+ designation.
Brad – to your question about what to do if the camera lens gets dirty – the answer is a little bit of manual labor. While the lens covering is made of special material that doesn’t bind well to dust and dirt, sometimes a car drives through a lot of mud and material gets caked onto a lens.
When this happens, the car will notify the driver of the visual impairment so that we’ll know to clean it off. I’ve actually had this happen a few times with my own car. All it took was a quick wipe and the camera lens was clean and functioning normally.
As for driving in inclement weather of various kinds – snow, fog, heavy rain, and so forth – this is still one of the challenges autonomous technology is working on.
Snow, for example, poses a number of unique challenges, such as obscuring our vision of the road and other objects, increasing glare and reflections, and reducing tire traction.
Snow driving also requires different techniques – we limit our speed more than usual, avoid lots of stopping and starting, and must be extra cautious when going up and down snowy or icy hills, as just a few examples.
That’s why having a large database to pull from is so important in this field. And Tesla currently leads the pack in this area, with over 5 billion miles recorded on Autopilot as of the start of this year. This is one of Tesla’s big advantages over the competition.
Right now, Teslas still require driver awareness and readiness to intervene. But as more and more Tesla drivers practice maneuvering in all kinds of complex conditions, we’ll continue to see the kind of rapid improvements in Tesla’s self-driving technology that we’ve seen over the last few years.
The AI behind Tesla’s system continues to learn how to respond in snow, fog, or other challenging situations as its Autopilot program becomes more experienced.
And once we accomplish that, we will truly be on the path toward Level 5 self-driving cars, capable of taking us anywhere in any kind of conditions.
My prediction is that we’re not far away at all. This is a near future event. And I believe that most consumers are going to be shocked at what Tesla’s cars will be able to do by the end of this year.
And I for one can’t wait.
In the meantime, if any readers would like to learn how to invest in the future of transportation and the self-driving trend, go right here for more details.
That’s all we have time for this week. If you have a question for a future mailbag, you can send it to me right here.
Have a good weekend.
Regards,
Jeff Brown
Editor, The Bleeding Edge
Like what you’re reading? Send your thoughts to feedback@brownstoneresearch.com.
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
