Hi, I’m Azeem Azhar. I convene Exponential View to help us understand how our societies and political economy will change under the force of rapidly accelerating technologies.

Earlier this week I wrote about big tech’s PR problem. 

💎 In my latest podcast, I tackle deep tech investing with the seasoned investor Matt Ocko, co-Managing Partner and co-founder of DCVC. We go into what it takes to found companies and invest in deep tech space, and why Matt thinks deep tech could bring about the age of abundance.

Dept of the near future

💾 The Fraunhofer Institute appraised the value of opensource in the EU at €63 billion per year, or 0.4% of GDP. Marginal investments could easily take that to €100bn per annum. 260,000 of the 3m software developers in the bloc (which included the UK at the time of the research) are involved in opensource. Every full-time equivalent employee developing opensource generates four times more GDP. What is working in Europe, is working on Mars: “This the first time we’ll be flying Linux on Mars. We’re actually running on a Linux operating system.”

🏢 Open source depends on digital collaboration. Urban agglomeration is one enabler of collaboration in the physical world. We know that innovation takes place in cities at much higher rates than in rural communities. Matt Clancy gets to the heart of why cities have this effect. When people are close together, introductions and meetings take place. Once people initiate and consolidate new relationships, being physically close is no longer as important. 

💡 The models that drive AI systems are the factories of the future. Nvidia’s Bryan Cantazaro reckons that by 2025 the most sophisticated language models will cost $1bn to train. Bear in mind that on a price-performance basis, compute in 2025 will be five to 20 times cheaper than today. So in terms of today’s compute, that would be $5 to 20bn dollars. OpenAI’s GPT-3 was rumoured to cost $12m to train last summer, although Sam Altman, OpenAI’s boss, told me it was much more. Interesting commentary from James Wang, reflecting on his forecasts for training scale:

James Wang @draecomino

The original forecast we did for AI training landed at $1 billion to train the largest model by 2025. But it seemed so crazy that we plotted a range and landed at $100m. Nvidia's Bryan Catanzaro @ctnzr's just said he could see $1B models by 2025. Lesson: follow your conviction.

2:47 PM ∙ Feb 17, 2021

---

237Likes43Retweets

The amazing thing is that we can start to compare the cost of training single AI models with the cost of building the physical fabs that make chips. TSMC’s state-of-the-art 3nm fab will run to around $20bn when it is completed in two years. A fab like this may be competitive for 5-7 years, which means it’ll need to churn out $7-8m worth of chips every day before it pays back.

🧑‍💻 Spotify says that employees can work remotely from anywhere and still earn a San Francisco salary. It’s critical to think about the downstream effects of these decisions. The leader of the Western Cape province in South Africa announced a push for a remote working visa targeting digital nomads to entice them to work from Cape Town. Given the cost of living and the easy time zone, places like Cape Town are attractive choices for employees freshly unchained from their offices in London or New York. 

Dept of Corona

💉 Covid-19 is here to stay. With vaccine deployment well underway in the West (but slowly making its way to emerging markets), the challenge is how to mitigate and control the virus. Will we need to segregate society based on who has been vaccinated and who hasn’t? This issue is already salient in Israel, thanks to the country’s rapid vaccination drive. Senior health officials are pushing for a tiered system whereby those with vaccinations can go to restaurants, attend conferences, and use gyms. Those without the jabs can’t. This is fascinating ethical territory. The Ada Lovelace Institute, where I am on the board, has explored the ethical issues of digital vaccine passports.

🚑 In the US, Covid-19 has taken such a large toll on society that life expectancy must be revised downward by 1.13 years, hitting Black and Latino groups hardest. Scientists are rushing to identify them before they get out of control: a sort of weather forecasting system for emerging diseases.

🔋 Dept of decarbonisation: 415.78 ppm | 3,398 days

Each week, I’m going to remind you of the CO2 levels in the atmosphere and the number of days until we reach the 450ppm threshold.

The latest measurement (as of February 19, 2021): 415.78 ppm; January 2020: 412.37 ppm; 25 years ago: 360 ppm; 250 years ago, est: 250 ppm. Share this reminder with your community by forwarding this email or tweeting this.

🚙 QuantumScape, an electric vehicle battery startup, figured out a way to produce multilayer battery cells, which will be crucial for pioneering solid-state lithium-metal batteries for electric cars. (QuantumScape went public via a SPAC last year, currently valued at $22bn, it is still more than 50% its frothy peak a couple of months ago.)

Short morsels to appear smart during your next power outage

🚀 Filecoin, the co-operative open distributed file storage system which uses the Filecoin token as an incentive mechanism, now has a global capacity of 2.5bn gigabytes. 

👨‍🎨 The artworld is also opening up to the blockchain. Christie’s will be the first auction house to offer a standalone non-fungible token—for a digital work by Beeple. 

🌎 Applying the pandemic mindset to larger challenges such as climate change. Enter planetary realism. 

💻 That was fast. Chrome OS overtakes MacOS as the world’s second most popular desktop operating system on Earth. (Linux is the most popular OS on Mars.) 

👀 Most agree that the term Artificial Intelligence was codified at a Dartmouth workshop in 1956. At a conference five years earlier, however, the concepts were discussed in detail, just not named. 

🇬🇧 Britain is poised to have its own version of Darpa called Aria with a GBP800m budget. (See also, my recent discussion on how to make innovation happen with Darpa’s former Director Regina Dugan.) 

📚 It’s all about appearances. Adding a bookshelf changes the results of AI-based job video interviews looking for personality traits. 

💻 The Solarwinds hack was one of the “largest and most sophisticated” ever. 

🎻 A terrible user interface cost Citi $500m.

🛸 Fascinating. Declassified Soviet spacecraft and orbital weapons. This piece is from 2018 but still makes for great reading. 

🧘‍♀️ Growth Hacking comes to Clubhouse in the form of “silent” rooms from Egypt to Silicon Valley.

Endnote

One key pillar of the Exponential View thesis is the expanding tension between the potentials of fast-moving technologies and the pace of institutional changes. And that, in order to deal with that tension, we need to establish new rules for the road.

Two existence proofs of the thesis this week. The first was the kerfuffle between Facebook and Australia. The second is that the UK Supreme Court decided in favour of two Uber drivers, rejecting a proposal from the ridesharing firm that the drivers were self-employed entrepreneurs. (This thread is worth reading on that case.)

Both of these cases exhibit the tension that emerges during this period of societal change. My take on the proposed Australian law, the News Media and Digital Platforms Mandatory Bargaining Code, is that it is hopeless and muddle-headed, but if it becomes law firms must adhere. If it does end up on the Australian statute books, it will represent a failure of the legislature to promulgate a good law. Rather than increase competition, it will cement the oligarchy of digital platforms and media providers.

To understand why read EV reader, Joshua Gans, who has advised many governments and regulators, including the Australian Competition Commission.

The entire process has the surely unintended consequence of enhancing the very market power that it was supposedly concerned about.

There is no improvement in any competitive outcome whatsoever. It is the codification of an oligarchy. For those outside, notably Australian consumers, it offers nothing.

Legislative incompetence also marks the Uber case in the UK. The Supreme Court has had to step in and make do with existing statute and case law, which pre-date gig working. The court found Uber breached them.

Parliament needs to step up and actually restate our employment law to reflect modern working arrangements. This restatement may still bar Uber’s existing practices but it’s critical that it be based on modern, fit-for-purpose legislation. Gig working mediated through digital platforms has been a thing for nearly a decade, it really is time to step up.

Cheers,

😜 Azeem

P.S. If you enjoyed this issue, please take a moment to forward it to ONE friend. Cheers!

What you’re up to – notes from EV readers
Achim Steiner, the head of the UNDP, published his TED talk on humanity’s planet-sharing powers and what they mean for the future. 

Josh Hardman’s Noetic Fund is about to open its second Fund, which is devoted to investing in early-stage psychedelic medicine companies. 

Peter Scott will be delivering the opening keynote at the British Telecom AI Festival on February 24.

Cedric Vanleenhove published a book with Jan de Bruyne called Artificial Intelligence and the Law. 

Kenneth Cukier published a commentary on "How AI Shapes Consumer Experiences and Expectations" in a special issue of the Journal of Marketing. 

Nithya Sambasivan published “Re-imagining Algorithmic Fairness in India and Beyond”. 

Denis Rivin’s six short videos on the theme of 'Radical advancements to keep an eye on in 2021' in collaboration with the Nordic hub for sustainable urbanization, BLOXHUB.

To share your projects and updates, fill out your details here. Because of space constraints, we prioritize updates from paying members and startups I have invested in. (You can become the former by subscribing, if you have not already, and the latter by getting an intro to me via a trusted contact.)