Today’s Weekly Dose is brought to you by… Baseten

Slow models = dead products.

Fifteen years ago, Amazon found that every 100 milliseconds of latency cost 1% of revenue. Today, 100 milliseconds kills entire products.

That's why startups building real AI-native applications, not GPT wrappers, turn to Baseten for applied performance research, distributed GPU infrastructure, and hands-on support.

Leading generative AI companies like (Not Boring Capital portfolio company) Writer, Bland, and Rime choose Baseten for mission-critical AI inference workloads where latency and reliability are both crucial.

With Baseten, you get production-ready deployments for any open-source, fine-tuned, or completely custom AI model, and Not Boring readers can get up to $250 in free credits to test the platform.

Learn About Baseten

Hi friends 👋,

Happy Friday and welcome back to our 121st Weekly Dose of Optimism. Bit of a return to normal this week: no election buildup and less election fallout. And a normal week is generally a pretty good week, if you’re paying attention. History-making rocket launches, breakthroughs in quantum mechanics, drug discoveries, and, yes, a dose of optimistic political news.

Let’s get to it.

(1) SpaceX Starship launches banana to space, skips giant rocket catch on 6th test flight

Mike Wall for Space

Today's mission aimed to do far more than just bring Super Heavy back to Earth in one piece. SpaceX also wanted to put Starship's upper stage — a 165-foot-tall (50 m) spacecraft called Starship, or simply "Ship" — through its paces.

It was just another day at SpaceX this week, when the company successfully completed its sixth launch of Starship. While flight data did not indicate a chopstick-catch landing was possible, the launch was still successful in all other regards.

The mission successfully tested Starship's reentry capabilities, including modifications to its heat shield and engine relight in microgravity, while also achieving a controlled splashdown of both stages. These advancements are critical for developing a fully reusable spacecraft capable of supporting lunar landings, Mars missions, and whatever other civilization-altering missions the company has planned.

It’s amazing that the company has turned launching and safely landing Starships into an almost common occurrence in such a short amount of time. As a reminder, Starship is over 400 feet tall, is the largest and most powerful rocket ever built, and dwarfs the Statue of Liberty. We can now launch the Statue of Liberty into space and land it safely back on earth, and it’s not even a big news story anymore.

Elon is on an all-time heater. We simply must just let this man cook.

(2) Javier Milei: President of Argentina - Freedom, Economics, and Corruption | Lex Fridman Podcast #453

Lex Fridman Podcast

Speaking of just letting people cook, President Javier Milei went on Lex’s podcast this week and explained just what he’s been up to in Argentina over the past year. His aggressive austerity measures have delivered astounding results: achieving a fiscal surplus for the first time in over a decade and reducing monthly inflation to below 3%, the lowest in three years. His administration has already removed or significantly amended over 300 regulations, and now alters around five regulations per day with the goal of making changes to over 4,000 laws. You may not like his style, but the results speak for themselves.

You may be asking yourself, “Why should I care about what the President of Argentina is up to?” The answer is pretty clear: Milei’s approach to government efficiency and deregulation is serving as the playbook for the US Government’s forthcoming Department of Government Efficiency (DOGE.) Don’t take my word for it — both Musk and Milei are on record stating as much. Yesterday, Elon tweeted a video of Milton Friedman doing the video version of Leonard Read’s I, Pencil.

Beyond our shared enthusiasm for governmental efficiency, three other things stood out to me from this conversation.

1. It’s super refreshing to hear from a politician that is extremely well-versed in economics and doesn’t talk strictly in platitudes when it comes to the economy. Even if you don’t agree with all of his conclusions, you’d likely agree that he is well read.

2. Milei does about as good of a job as anyone explaining the origins and dangers of wokeness. His analysis is not inflammatory and doesn’t simply cherry pick examples from ‘Libs of TikTok’, but rather, it is rooted in a deep understanding of the history and philosophy of socialism.

3. ElevenLabs is sweet. The original conversation occurred between an English speaking Lex, a Spanish speaking Milei, and an interpreter. But the podcast was presented using an overdub from ElevenLabs that made it sound like a fully English conversation between Lex and Milei. It was a very high-quality overdub and, if Lex had not broken it down at the beginning of the podcast, I am not sure I would’ve picked up that Milei’s voice was AI-generated.

(3) A mechanical qubit

Yang et al in Science

By coupling a bulk acoustic wave resonator to a superconducting qubit, the authors engineered a parameter regime in which a mechanical mode of the resonator inherits the nonlinear features of the superconducting qubit necessary for strong phonon-phonon interactions. They then demonstrated a full set of single-qubit operations on their mechanical qubit.

Researchers have created a new type of qubit using mechanical vibrations, called a mechanical qubit, by coupling a resonator to a superconducting qubit. As a reminder, a qubit is the fundamental unit of quantum information, capable of existing in multiple states simultaneously (superposition) for advanced computing and information processing.

The mechanical qubit allowed them to engineer strong interactions between phonons (quantized sound waves) at the single-quantum level, something previously impossible in mechanical systems. They demonstrated full control of the qubit, including initialization and quantum gates, with the system showing remarkably long quantum lifetimes.

If you didn’t understand that, you’re not alone — here’s why it matters: the breakthrough opens the door to using quantum sound for advanced technologies like highly sensitive sensors, powerful simulations, and efficient quantum computers. It’s a leap toward making quantum acoustics a key player in the future of quantum technology.

Separately, researchers showed that AI outperforms humans in quantum error correction.

Progress in quantum is accelerating. It feels like this is another “always-a-decade-away” technology that might come sooner than most people realize.

(4) New schizophrenia drug could treat Alzheimer’s disease

Diana Kwon for Nature

Since a schizophrenia drug, the first in decades with an innovative mechanism of action, gained US regulatory approval in September, some researchers have proclaimed a new era for psychiatric medicine. About half a dozen similar drugs — for schizophrenia, Alzheimer’s disease and other conditions involving the brain — are in various stages of development, most in early-stage clinical trials.

Back in September, the FDA approved a new schizophrenia drug, KarXT (Cobenfy). The approval represented a significant innovation in psychiatric medicine by targeting muscarinic receptors, a mechanism distinct from traditional dopamine-based drugs. Muscarinic receptors are proteins in the brain and body that respond to the neurotransmitter acetylcholine and play a key role in cognition, movement, and other physiological functions. KarXT approval for schizophrenia has renewed interest in targeting muscarinic receptors for other brain conditions, such as Alzheimer’s.

Targeting muscarinic receptors, particularly the M1 subtype, could help treat Alzheimer’s by enhancing acetylcholine signaling, which is crucial for learning and memory. Activation of M1 receptors has been shown to improve cognitive function and may slow neurodegeneration by supporting brain cell communication and reducing the buildup of harmful proteins like beta-amyloid and tau. Importantly, this approach would not only treat existing symptoms of Alzheimer’s but may also address the underlying diseases.

We’re still pretty long ways away, though. Other muscarinic receptors drugs have failed to pass through trials. The challenge is due to the difficulty of selectively targeting specific receptor subtypes (like M1) without causing widespread side effects, as these receptors are found throughout the body. That said, the FDA’s approval KarXT has revitalized confidence in muscarinic receptor-targeting drugs and their potential for treating a wide variety of brain-related conditions.

(5) Gary Gensler Announces Resignation from SEC

On January 20, 2025 I will be stepping down as @SECGov Chair.

Packy here. Whew. We knew this was going to happen the second Trump was elected, but it just feels good to hear him say it, ya know?

Gensler’s four-year tenure atop the SEC was defined by his policy of regulation by enforcement — establishing rules through lawsuits and penalties instead of guidance — and overreach. Fittingly, just yesterday, a US District Court struck down the SEC’s Dealer Rule, concluding that “the SEC exceeded its statutory authority.”

That all of this is happening while prices soar — Bitcoin is approaching $100k, Solana just hit an all-time high — isn’t lost on anyone. But “number go up,” while very fun, isn’t necessarily optimistic.

What I’m most excited about is that, between Gensler’s resignation, positive momentum behind clear crypto regulation in Congress, and price movement, the a16z crypto price-innovation cycle will be able to spin smoothly.

Price —> interest —> developer activity —> companies —> great products.

For the past four years, the cycle got stuck somewhere between companies and great products. Developers got excited, only to discover that the real version of what they were trying to build would land them in court. Now, coupled with real improvements in infrastructure (stablecoins, < 1 cent transaction fees, account abstraction, etc…), I expect to see the price excitement translate into the most compelling suite of crypto products we’ve gotten yet.

Crypto is already proving its usefulness. Bitcoin is a store of value. Stablecoins are better money infrastructure, as evidenced by Stripe’s $1.1 billion acquisition of Bridge. Polymarket played a significant role in the 2024 US Presidential election (and got it more right than the media and pollsters). The idea that there are no use cases is dead. Now, the question is how many really great products will be built this cycle.

As I wrote in Blockchains as Platforms, blockchains are platforms with unique benefits and real drawbacks, both technical and regulatory. “Whether a developer chooses to build their product onchain depends entirely on whether the benefits outweigh the drawbacks for the specific product they’re trying to build.” As those drawbacks decrease, as building useful things becomes less likely to land you in jail, more developers will choose to incorporate crypto’s unique benefits into their products.

Bubbles are good. Done right, they concentrate talent and resources on big problems and pull the future forward. Gensler was a kink in the bubble machine. I’m excited to see what happens now that the machine will be allowed to run smoothly.

Bonus: AI helps uncover metals in Australia critical for clean energy

Andrew Freedman for Axios

At about 2 kilometers by 2 kilometers, or 1.2 miles by 1.2 miles, Earth AI says this expansion could make it one of the largest platinum group elements prospects globally.

Packy again with an update on our portfolio company, Earth AI.

Since I wrote about the company in April, they’ve continued to drill out their discoveries to validate their size and viability. Yesterday, along with their partner Legacy Minerals, Earth AI announced that the palladium-platinum discovery at Fontenoy might be very big. In mining speak: “Assay results from the second phase of drilling received have confirmed a large mineral system.” In English: the thickest hit was 360 meters deep, as deep as the Empire State Building is tall, and the mineralized area is 2 kilometers by 2 kilometers wide. Fontenoy has the potential to become one of the largest projects of its kind on the planet.

As an investor, this is exciting news, because if further drills continue to confirm the value of the deposit, it could allow Earth AI to spin its flywheel: sell stakes in the project, put that money into more rigs, drill out more discoveries faster and cheaper. For the world, it’s exciting, because Russia and South Africa are the world’s two largest producers of palladium, a critical energy transition metal, and Russian exports may face sanctions. Finding and mining the metals we need in allied countries is a W.

Have a great weekend y’all.

Thanks to Baseten for sponsoring! We’ll be back in your inbox on Tuesday.

Thanks for reading,

Packy + Dan