It’s been a busy week…
Thank you all for your patience as I’ve been working on this week’s AMA. I had a very busy travel week and disruptions caused by delayed flights that bumped me off my normal schedule. Rather than try to rush out my AMA, I decided to take my time and do it right on Saturday.
But before we get into today’s issue, I want to take a moment to invite you to join me for a sit-down on Wednesday morning.
I’ll be sitting down with a colleague and friend of mine – Wall Street legend Whitney Tilson – to discuss something that’s been brewing in the semiconductor industry…
A little-known company that’s developed an artificial intelligence “super chip” that’s so powerful, it’s one of very, very few companies that I would consider a worthy competitor to the market darling of artificial intelligence – NVIDIA.
There are precious few tech companies that can even stand close to NVIDIA’s industry dominance, let alone with the distinction of developing a tech powerful enough to potentially challenge that.
I’ve had my eye on this company for years – since long before a handful of industry experts started whispering about it – and I believe this super chip could very well establish a new order in the stock market.
As powerful as the Magnificent 7 are… I’ve also got my eye on five companies that could dethrone them once this super chip starts making waves.
I don’t want to take up too much of your time away from the AMA, but if you’d like to learn more, you can go here to sign up to join me and Whitney on Wednesday, May 21, at 10 a.m.
I’ll get into the incredible work being done at the company behind this chip, as well as the handful of companies I believe are going to soar once this chip goes mainstream in the markets… which I believe could happen as soon as early June.
Make sure you go here to add your name to the guest list, then read on for today’s AMA…
Hi Jeff, as always, eager to read your highly useful insights in the techno world.
This comment relates to your previous service with the other publisher and some recommendations of BioTech companies that, back then, were achieving great advances in the field of genome modification.
As you have recently expressed, we are in a BioTech winter, and consequently, their stocks have been badly beaten. However, I understand that you still believe in a good future for them. My concern is that with the advent of AI, most of the old and new companies working in genome modifications will have the opportunity to level the playing field and reduce any advantage that those past recommended companies may have had two or three years ago.
Based on the above, I wonder whether there is any real chance for companies like Editas, Cybin, BioMarin, etc., to again lead the race and have their value rise again.
May we have your thoughts on this issue? I mean, the impact of AI in genome modification companies as a whole? What specific features would be key for a company in this field to lead the race? And do you expect to again make specific recommendations, let’s say, within 2025?
Thanks a lot.
– Alejandro G.
Hello Alejandro,
What an insightful question. And it reveals one of the most interesting market impacts of AI on the biopharmaceutical industry. And there is a unique question concerning market structure that has yet to be answered regarding the adoption and utilization of this powerful technology.
But to answer your question about whether or not existing companies will be impacted negatively by new companies using AI, the answer is quite simple.
They’ll be on even ground if they adopt artificial intelligence and, in some ways, have an advantage because their database of knowledge on whatever diseases they have been working on will typically be of higher quality than that of a new entrant.
But those companies that don’t leverage the technology, to your point, will be at a serious disadvantage.
Now, what the industry is currently trying to figure out is how to employ these frontier AI models. Should they hire a large internal computer science team at great cost and build an AI model from the ground up, train it, and then utilize it for drug discovery? Or should they partner with an AI company from which they can license the AI model for the same purpose?
Ironically, most of the biotech industry has been leaning towards trying to raise massive amounts of capital, hire an in-house AI team, build from scratch, and hope to leverage AI for their drug.
While this is common, it is highly inefficient, far more expensive, and actually takes more time.
The other model is exactly the opposite. There are a handful of AI companies that specialize in providing AI tools to support drug discovery that license access to their technology.
An example of a company like this is Burlington, Vermont-based Syntensor, which is still a private company.
There’s a great analogy to the semiconductor design space. Semiconductor companies for decades tried to design their own software in-house that they’d then use to design semiconductors, but developing software wasn’t their core competency. Designing and manufacturing semiconductors is what they were good at.
That industry evolved to have two large players – Cadence Design Systems (CDNS) and Synopsys (SNPS) – which specialize in software for designing and manufacturing semiconductors. Everyone in the industry works with one or both of these companies.
Comparatively, for biotech, many companies will be better served utilizing AI models and applications from existing AI software companies focused on the life sciences industry, like Syntensor, for their own technologies and research than they will be developing – at higher cost and less efficiency – their own in-house AI models.
But one thing is for certain: the companies that fail to adopt or utilize AI at all will be at a material disadvantage to those that already have or are already working towards employing AI in their research and development process.
And yes, I do believe that we’ll finally be able to provide coverage and research on the biotech industry before the end of this year.
I continue to work on this space and am doing a lot of research in the background. The private capital is already flowing into biotech again, which is very positive.
But the IPO market has not yet come to life, and Powell continues to play dangerous political games, maintaining Fed Funds rates at elevated levels that directly impact small caps and the biotech market.
I do expect to see some positive developments on this front within the next three months. Inflation hasn’t been as low as it is now since early 2021. The data strongly supports lowering rates. Any further stalling at this point is just Powell prolonging this game of political chicken with the U.S. government to the detriment of the entire country.
More to come.
Hey Jeff,
I am an unlimited subscriber and have been for some number of years. As others have commented, I greatly appreciate everything you do and look forward to all of your content.
That said, I am about halfway through reading your response to this question — Human-Supplied Data Deterioration in the Long Term? — and I am troubled. I don’t intend for this to be a political comment, but it seems unavoidable.
Filters are applied to training data. The goal is to remove low-quality data as a training input and apply a higher weight to higher-quality data for training. As an example, I can’t help but think about all the “DEI” initiatives that are currently being implemented. There was a news item today about all books in all libraries were being reviewed for “DEI content” and would be removed. Would such books be considered “low-quality data” and therefore be filtered out? Or, would they be considered so that both sides of the narrative could be presented in a trade-off type of scenario? It seems like, in today’s environment — one man’s “truth” is another’s “low-quality data.”
How will AGI/ASI remain neutral and not become increasingly biased one way or the other?
Looking forward to your response,
– Les L.
Thanks Les,
You’ve hit on a heavy and very difficult question.
And it speaks to a trend that has been in motion for the last couple of decades, propagated by a political ideology that humans, particularly those in academia, should decide what we should learn and what we should think.
This stands in stark contrast to the original purpose of education, which was to teach how to think (not what to think), as well as how to solve problems, think critically, and identify mistruths.
AGI/ASI can remain neutral if the governing body of the AGI/ASI – or the executive body in a company that has developed AGI/ASI – uses a framework for training data limited to facts, truths, and evidence-based data.
DEI is a human construct developed by those with a political ideology. A book on DEI could still be an input to an AGI (i.e., it could be read by an AI), and it would be analyzed and deconstructed as political ideology and narrative, not as a governing framework.
Until we see it, I imagine it will be hard to picture just how significant a leap in performance AGI will have over the AI models many of us have grown used to interacting with. AGI – especially ones developed by companies that are transparent about how they train them and their motivations for doing so – will be far more advanced and far more discerning than the average chatbot.
So, to answer your question more specifically, yes, such kinds of books or information would be considered, but they would be presented in a factual light for what they are, not simply consumed and wholly accepted as fact.
My comments do assume that whoever is governing the AGI/ASI is using a maximum truth-seeking framework for its AGI training and development. Today, only one company is taking that approach – xAI.
All others are programming certain degrees of bias into their frontier AI models. This is both dangerous and irresponsible.
The bigger battle will actually be the battle for the most widespread use of a neutral AGI/ASI for global society. This is a battle even bigger than the push for an Orwellian, totalitarian government in Western society, which has lost, at least for now, in the U.S., but it is still being pushed in Canada, Australia, New Zealand, the U.K., and most of Western Europe.
A biased AGI/ASI will be a powerful tool for those who despise personal freedoms and thirst for total control over their populations.
Elon Musk understands this deeply. It’s the only reason he acquired Twitter, and it’s also the reason that he built xAI to develop a maximum truth-seeking AGI, and ultimately an ASI.
He knows the stakes are high, and that is why he’s all in on this mission… and why I believe he will be the first to achieve AGI. And at the pace he’s going, I’ve predicted we’ll see that within a year.
This is the biggest battle the human race has ever fought in history. And it’s all happening right now. We’re seeing it happen in real time.
And it’s up to us as free, sovereign individuals to choose which future we want for ourselves and our families.
Dear Jeff,
I have been a fan of Thorium/molten salts reactors for a long time. Admiral Rickover, in his development of the nuclear Navy, refused to consider them because they would not produce weapons-grade radioactive material. Is there a program extant that is developing compact reactors that could power small communities and large AI installations?
– Timothy R.
Hi Timothy,
You are spot-on. The only reason that the U.S. didn’t choose thorium as an approach for nuclear fission was that it would not produce weapons-grade radioactive material.
The choice of uranium was due to the wartime advantages that a uranium energy focus would have for the U.S. military and defense.
I wrote about this in another recent AMA, The Bleeding Edge – Autonomous Ride-Hailing in Rural Areas…
It is true that Uranium 233 can be used to produce a nuclear weapon. And it is also true that Uranium 233, which is bred from a thorium reactor, is less effective than Uranium 235 for a nuclear weapon.
The reality is that when the U.S. government was researching fuels and nuclear reactors to meet the exponential growth of energy demands of the U.S. economy, it found that uranium-based nuclear reactors were simply more efficient than thorium-based reactors. This was the primary driver for the industry, leaning heavily into uranium versus thorium as a fuel.
That decision was at least partially influenced by the proliferation of nuclear weapons at the time and the need for enriched uranium for that purpose.
It’s unfortunate, but I’m certain other countries would have pursued uranium as a fuel, and ultimately nuclear weapons, regardless. And at this stage, all of the legacy nuclear power ecosystem is designed to support uranium, so we shouldn’t expect a pivot to thorium as a fuel.
Far more likely is a migration to small modular reactors (SMRs), as you’ve suggested, capable of powering small communities and large AI installations.
There’s not really a program extant per se, but there is an industry movement with strong government interest to support such a path.
In 2020, the U.S. Nuclear Regulatory Commission (NRC) approved the first SMR design by NuScale Power (SMR). TerraPower, X-Energy, and Oklo are all working on similar small reactor designs that can be used for the purposes you suggested.
The SMR approaches by the companies mentioned above are all evolutions of the last generation of nuclear fission technologies. We can think of these companies as working on Gen-4 nuclear fission technology.
And the real revolution in clean energy is happening right now in nuclear fusion, which is an entirely new industry and technology of its own.
Fusion reactors are perfectly designed for the applications that you suggested, and some are even portable. One of the coolest concepts that I’ve seen in fusion is from Avalanche Energy and its Orbitron.
The Orbitron | Source: Avalanche Energy
Either way, these smaller solutions for clean energy are being built quickly out of necessity. Which is actually a good thing.
It is no longer necessary to have a taxpayer-funded program for this development. The private markets have stepped up to fund the development of this technology, and it is happening right now.
Both SMR prototypes and fusion prototypes exist today and are nearing the point of being commercialized in the next two to three years.
Very cheap, clean, and abundant energy for all.
We have so much to look forward to.
Hey Jeff,
I’m super excited about Tesla’s robotaxi rollout next month, but I’m also curious about some of the company’s other innovations. Primarily, I am wondering what kind of work they are doing with their solar rooftop and Powerwall products. I remember you writing about them when they first came out, and I think they had a huge waiting list. Do you or anyone you know have them? If so, what do you think?
Thanks,
– Synthya G.
Hi Synthya,
Yes, you’re right. Tesla has been making a big push with its solar rooftop and Powerwall products.
These are so complementary to Tesla’s electric vehicle. The idea is that we can use electricity generated from solar panels on our roof to power our EVs. That is an ideal design, as most EVs are powered by electricity created using fossil fuels (which defeats the point).
But Tesla has basically put its solar tile project on the back burner, for now. This is really a shame, as this is a product with such a long waiting list.
Tesla does still manufacture a limited number of solar tiles for specific markets out of its Buffalo, New York, gigafactory, but it is not a priority business.
Tesla continues to have difficulties getting the cost of manufacturing low enough on its solar tiles so they make economic sense for buyers. This problem doesn’t exist with standard solar panels as they are much easier and cheaper to manufacture.
And when people sign up for solar products from Tesla, these solar panels are not manufactured by Tesla. Tesla sources private-label, Tesla-branded solar panels installed by third-party installers.
What Tesla does focus on is its own battery technology and power storage through its Powerwall and Megapack.
In 2024, Tesla’s Energy Generation and Storage group made up about 10% of Tesla’s total revenues. If I had to guess, only about 2–3% of that was related to solar panels, and definitely a fraction of that related to solar tiles. The majority of those revenues are from Powerwall and Megapack products, which makes perfect sense.
For homes, the combination of solar panels and Powerwalls is fantastic. Not only is it a great solution for powering an EV, but it can also act as a backup for the house should utility-supplied electricity go down due to a storm or a line being cut.
That’s everything for today’s AMA. As always, you can reach us with questions or comments through our feedback channel right here. Let us know what you think.
I hope everyone enjoys the rest of their weekend.
Jeff
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.
