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

Speakeasy recently raised $15M to revolutionize REST API development.

Even the most gifted craftsman needs the appropriate tools to build something they’re proud of. Unfortunately, the API tooling ecosystem is still stuck in 2005.

Speakeasy wants to make it trivially easy for every developer to build great APIs. To make that dream a reality, Speakeasy is building a platform to handle the heavy lifting of API development and unburden developers to focus on refining the business logic.

From fast-growing innovators like Vercel, Mistral AI, and Kong to hobby projects by solo devs, engineers use Speakeasy to accelerate their API development and adoption.

Learn About Speakeasy

Hi friends 👋,

Happy Friday and welcome back to our 120th Weekly Dose of Optimism. Even during the most chaotic of times, we can rely on the world producing (at least) five optimistic stories worthy of coverage in The Weekly Dose of Optimism. It’s like clockwork. This week, Joe snuck in some huge last minute plans for a nuclear buildout, plus the future of nanotechnology, distributed networks, gene editing, and much more. In fact, this might be the most action-packed Dose in a long time. Humans rock.

Let’s get to it.

(1) US Unveils Plan to Triple Nuclear Power by 2050 as Demand Soars

Under a road map being unveiled Tuesday, the US would deploy an additional 200 gigawatts of nuclear energy capacity by mid-century through the construction of new reactors, plant restarts and upgrades to existing facilities. In the short term, the White House aims to have 35 gigawatts of new capacity operating in just over a decade.

Go Joe!

On Thursday, the Biden Administration unveiled plans to triple US nuclear power capacity by 2050, adding 200 gigawatts through new reactors, plant restarts, and upgrades. In addition, the plan calls for 35 gigawatts of new capacity within a decade.

This is one of the most ambitious plans for nuclear energy ever laid out by a U.S. president — instead of maintaining the status quo or focusing entirely on safety concerns, the plan meets bipartisan momentum for nuclear energy, which has been pulled forward by the AI boom and other geopolitical issues. Nuclear buildout has stagnated for the last 50+ years, with barely any capacity added since the 90s. This plan flips that with a goal to not only increase nuclear capacity, but triple it within 25 years.

The plan itself is pretty on point. Joe must’ve been a Age of Miracles listener. The plan calls for capacity expansion through the construction of new sites and the revival of idle plants, the deployment of small reactors, and public/private partnership to make it all happen. The plan also calls for a number of regulatory and policy changes to make this all possible: expanding the skilled-labor workforce to build and work on these sites, removing regulatory barriers to construction and operation, and tax policies to make it all economically feasible.

Let’s just hope DOGE doesn’t come after this specific governmental initiative…

(2) Two Paths to Molecular Nanotechnology

Jacob Rintamaki

The creation of molecular nanotechnology (highly capable molecular machines that assemble products with atomic precision) could rival or surpass the impact of the Industrial Revolution. 

Our physical production capacity could be increased by many orders of magnitude while making our current computers, medicines, and machines look primitive by comparison. Nanotech, despite decades of infighting and setbacks, remains the world’s most underrated technology, and 

I believe the time to build nanotech is now.

Packy here. Jacob Rintamaki wrote a deep piece on what it would take to make molecular technology happen. Aside from being well-researched and ridden with rabbit holes to explore, this one is fascinating to me for a few reasons.

First, nanotech is one of those sci-fi technologies (like anti-gravity) that could decimate a large percentage of current companies when we figure it out, as we covered when we talked about J. Storrs Hall’s essay on Nanotech for Abundance Institute. It will be interesting to see what happens to the market if and when molecular nanotechnology approaches realization and scale.

Second, a few weeks ago, I wrote What Do You Do With an Idea?, in which I noted that many of the hard tech companies being built today are based on ideas from the 1950s and 1960s. The idea for manipulating matter at the level of individual atoms was proposed by Richard Feynman in a 1959 lecture, Plenty of Room at the Bottom.

In that essay, I wrote:

In addition to making new Einsteins  - genius researchers will always be important – we need to make new Vanderbilts and Fords and Wrights. Tinkerers who can pluck from the pile of withered ideas and combine them in new ways. We might also need new Librarians, who can help surface those great old ideas, and track which technologies and costs might make them economically viable after all of these years.

What I love about Jacob’s post is that he’s serving as the Librarian — taking the old idea, tracking its history, and detailing which technologies and costs might make it economically viable after all these years — and hinting in the conclusion that he might also be attempting to serve as the Vanderbilt, Ford, or Wright of Molecular Nanotechnology — plucking from the pile of withered ideas and combining them in new ways.

The list of things humanity might do with nanotech is practically limitless. In his piece, Hall previews step changes in manufacturing, robotics, construction, energy, and transportation, and a nail in the coffin of the idea that the world is resource-constrained.

I hope Jacob is right that the time to build nanotech is now, so I’m amplifying this: “if you are an investor, technologist (particularly in AI, protein design, or scanning probe microscopy), or just a nanotechnology enthusiast, then I would appreciate further communication with you via my X account @jacobrintamaki or my email address, jrin at stanford dot edu.”

(3) Evo: Creating Generative AI for Genomes

Arc Institute

By learning the information encoded within DNA using a frontier deep learning architecture, Evo is capable of both prediction and design not just at the level of DNA but also across RNA and proteins. Its interpretive and generative capabilities span biological scales, from nucleotides to the whole genome—bringing entire life forms into focus.

Jacob might have some help in the quest for molecular nanotechnology from Evo, a new foundation model for biology built by The Arc Institute, Stanford, and UC Berkeley researchers.

Evo, which appears on the cover of Science, can both interpret and generate biological sequences spanning from individual nucleotides all the way up to genome-scale designs over a million bases long. According to Arc, “Evo is capable of both prediction and design not just at the level of DNA but also across RNA and proteins. Its interpretive and generative capabilities span biological scales, from nucleotides to the whole genome—bringing entire life forms into focus.”

To demonstrate Evo's capabilities, the team designed EvoCas9-1, a fully functional CRISPR system that emerged after testing just eleven AI-generated designs. Despite sharing only 73% of its sequence with traditional CRISPR-Cas9, the new system achieved comparable activity - suggesting there may be vast unexplored spaces of effective biological tools waiting to be discovered. The researchers are now scaling up, moving beyond single-cell organisms toward more complex multicellular life forms, with the ambitious long-term goal of enabling an entirely new field of "genome design" where we could potentially create custom cellular pathways and organisms from scratch.

Evo is notable for it capabilities — it’s a tool that might one day help us design life — and for its origins — another win for the Arc Institute, and for new approaches to doing science.

(4) Google DeepMind open-sources AlphaFold 3

From VentureBeat

Google DeepMind has unexpectedly released the source code and model weights of AlphaFold 3 for academic use, marking a significant advance that could accelerate scientific discovery and drug development. The surprise announcement comes just weeks after the system’s creators, Demis Hassabis and John Jumper, were awarded the 2024 Nobel Prize in Chemistry for their work on protein structure prediction.

If you like using AI for biology, there’s more where that came from!

AlphaFold 3 is now open source! As a reminder, AlphaFold is an AI system that predicts protein structures, which is key to understanding how biology works. AlphaFold 3 takes it to the next level, modeling how proteins interact with DNA, RNA, and small molecules with incredible precision. It's a game-changer for drug discovery and molecular research. And it’s now available for academic use.

This is one of the positives of monopolies: Google has likely spent tens of billions of dollars building AlphaFold 3 and because it’s not directly crucial to its core monopoly business of search, it feels comfortable enough open sourcing it. The move will certainly accelerate scientific research and drug discovery. But it’s not entirely altruistic: the move bolster’s Google’s role as a leader in AI, builds goodwill with researchers, and ultimately should attract more top talent, plus the open sourcing is not without its limitations, as its only available for academic use and Google still protects its use in commercial applications.

(4) Mapping the ionosphere with millions of phones

Smith et al in Nature

Using smartphone measurements, we resolve features such as plasma bubbles over India and South America, solar-storm-enhanced density over North America and a mid-latitude ionospheric trough over Europe. We also show that the resulting ionosphere maps can improve location accuracy, which is our primary aim. This work demonstrates the potential of using a large distributed network of smartphones as a powerful scientific instrument for monitoring Earth.

I love when Silicon Valley becomes real life. Google just proved out a version of Pied Piper. Kinda. Sorta.

Google turned 40 million Android phones into a massive, crowdsourced scientific tool to map the Earth’s ionosphere in real time, filling gaps in satellite data. For the uninitiated, the ionosphere is a layer of Earth’s upper atmosphere filled with charged particles that affects radio waves and satellite signals.

This research enhances GPS accuracy globally, especially in previously low coverage zones, and could also unlock new insights on space weather. The coolest part, though, is that this was all achieved using everyday technology that was already in peoples’ pockets. Pied Piper may have not worked out (most startups fail!), but we’re happy to see that Google could build off of the technological breakthroughs the startup achieved.

Bonus 1: CRISPR builds a big tomato that’s actually sweet

Max Kozlov for Nature

Deleting just two genes that control sugar production makes a more succulent fruit.

*Imagine me delivering this story as David Friedberg in Science Corner*

Researchers used CRISPR to edit two genes in tomatoes, increasing their glucose and fructose levels by up to 30% without sacrificing size or yield. If you’re anything like me, your average grocery store tomato isn’t doing it for you — it’s watery and bland. This new methodology could restore the sweetness of tomatoes, without sacrificing size constraints or negatively impacting farming yields.

Of course, tomato sweetness is really just the start. It’s not a huge leap to imagine this technology being used to increase the flavor profile of a number of other fruits and vegetables, and importantly, doing so in a way that doesn’t negatively impact farming productivity or economics. Genetic modification of foods has gotten a bad wrap over the years — in reality, it’s a bit more nuanced: being able to mass produce safe, tasty food has come with tradeoffs. But this type of CRISP gene medication is generally viewed as better. It offers precise, natural-like genetic modifications without introducing foreign DNA, reducing unintended effects and improving safety and efficiency as compared to traditional GMOs.

Bonus 2: I tried to melt down a real-life nuclear reactor

Christian Keil for First Principles

Friend of Not Boring Christian Keil takes his show, First Principles, to Austin to pull the control rod out of a nuclear reactor and see what happens. Spoiler: he survived.

Bonus 3: The Return of Medici-Style Patronage

Packy again. On Wednesday, the legendary blogger Gwern went on Dwarkesh Podcast for his first public podcast, using an avatar and someone else’s voice to maintain anonymity.

During the conversation, he mentioned that he was living on $900-1,000 per month in Patreon fees, which is a shockingly low amount for someone so influential. Dwarkesh asked what it would take for him to move to SF, and he said $50-100k/year.

Dwarkesh set up a Stripe donation link, which I’m sure raised a ton of money, and then Mixpanel & Playground AI founder Suhail Doshi gave him the full amount.

It remains to be seen how Gwern’s writing holds up to being in the scene in SF — there’s something pristine about living the hermetic life in the middle of nowhere, with no obligations or temptations — but it’s one of the more public and generous recent examples of a resurgence of Medici-style patronage among people in tech.

Tyler Cowen does Emergent Ventures Grants. Jim O’Shaughnessy gives O’Shaughnessy Fellowships and runs Infinite Films. Marc Andreessen gave a $50k no-strings-attached grant to a twitter-based AI agent, Truth Terminal, and its creator, Andy Ayrey. Yesterday, the Collison-supported Arc Institute hit the cover of Science with its Evo model for predicting, generating, and engineering entire genomic sequences. Dwarkesh himself kept podcasting in the early days thanks to a $10,000 gift from Meter’s Anil Varanasi.

The more patronage of the arts and sciences, the better. Yet another sign that we’re living in a Renaissance.

Have a great weekend y’all.

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

Thanks for reading,

Packy + Dan