🔮 SVB; Metaverse reverse; energy storage; geopolitics of technology; baby Presidents ++ #413
Zuck's New Coke
Hi, I’m Azeem Azhar. As a global expert on exponential technologies, I advise governments, some of the world’s largest firms, and investors how to make sense of our exponential future. Every Sunday, I share my view on developments that I think you should know about in this newsletter.
In today’s edition:
Long-duration energy storage is the key to decarbonisation, but it’s been chronically underfunded,
Zuckerberg’s shoving the metaverse under the rug,
Food landscape of the future.
Sunday chart: Long storage, fast
$33 billion was invested in blockchain and crypto startups in 2021 alone. Only $1.8 billion was privately invested in Long Duration Energy Storage (LDES) startups between 2001 and 2021. LDES is a set of technologies that make it possible for power grids to have a large share of renewables, and be flexible enough to accommodate different power sources across weathers and seasons. They are designed to store energy from a few hours to weeks — much longer than the typical storage systems that are in use now.
The Long Duration Energy Storage Council’s lower estimate projects that we’ll need 1500 GW of installed power capacity of LDES. The biggest challenge for the moment, however, is not the installation, but the experimentation with the different competing technologies (thermal, electrochemical, mechanical, chemical) in the space. Similarly to other deep tech sectors, the risk profile of these startups makes them unattractive to many private investors.
To get the necessary funding to LDES startups, the Institute for Progress argues that “demand commitment mechanisms” driven by governments could jumpstart the necessary investment. I have been arguing for the last couple of months that governments are (rightly) starting to take a catalytic role, signalling strategic sectors of technology where markets are misaligned. I’m all for investing in LDES. We should also recognise that the storage problem is an intermittency problem. And we could address that in other ways. For example, high-voltage direct current interconnects could pipe electricity from sunny (or windy places) to where it is needed. Coupled with overcapacity this could be a natty approach. Virtual powerplants, knitting together car batteries, into mega grid storage could be helpful in medium-term timeframes. Fixing electricity markets so that 24/7 availability, which is expensive to provide, is treated as such could also help by creating the right incentives for investments.1
My weekly commentary: 😱 SVB & the tangled web
In my weekly commentary for premium members, I’ll quickly outline how the SVB calamity might play out. Watch your inboxes later today.
My key reads
Zuck’s New Coke: Zuckerberg is gradually, quietly escaping from the metaverse. Remember October 2021, when he announced that Facebook would become Meta, an umbrella company focused on bringing the metaverse to life. Now think back to 1985, when Coca-Cola changed the drink’s recipe to create “New Coke”, and discontinued it only a few months later, after it turned out to be a fiasco with consumers. We’re back to the future. This is Zuckerberg’s New Coke moment. Reality Labs, Meta’s the metaverse projects, has lost a total of $24 billion, more than half in the last year. Now, Chief Meta is jumping on the latest technology trends, while there is a quiet acknowledgement that his vision of the metaverse is on life support. The latest announcements include the release of LLaMA, the company’s own LLM. A few days later, The Platfomer revealed that Meta was working on a decentralised, interoperable Twitter alternative. The big surprise here: interoperability. Meta long resisted creating interoperable networks.
ChinAI: Twenty years ago, China was seen as not much more than the manufacturing capital of the world. A new report from an Australian think tank suggests that China has a global lead in 37 out of 44 strategic technologies, from defence, space, robotics, energy, biotechnology, AI and quantum tech. How did Beijing do this? Importing talent, deliberate design, and long-term policy planning. Shanghai in particular is positioning itself as the Chinese capital of AI, and has renewed the pledge to attract 20,000 to 30,000 AI stars to the city. Its strength lies in intelligent chip-design enterprises, but apparently lacks top algorithm talent. As I shared last month, China had a “Sputnik moment” with the reveal of ChatGPT, and developers are scrambling to release similar technology.
Russia’s Satellite Hack. At the beginning of Vladimir Putin’s invasion of Ukraine, Russia (most likely) managed a major cyberattack on a satellite that secures internet access across Europe. In this key read, Katrina Manson takes us through why the hack confirms Vladimir’s huge risk appetite, the attacks could have easily disrupted many European countries. The hack also exposes the vulnerability of our critical information infrastructure.
There are 5,000 active satellites circling the planet, and high-end estimates predict the number will top 100,000 by the end of the decade. While smaller, cheaper satellites are being launched into orbit by the thousands, the machines and the networks that run them remain woefully insecure.
Market data
Bing has surpassed 100m daily active users for the first time in its history. Thank Sydney for that!
The IDC forecast global spending on AI systems to reach $154 billion in 2023, up 26.9% from 2022.
Recent polls show that only 15% of Americans hold a favourable view of China, signalling a significant decline in public opinion. This shift in sentiment is underscored by the veiled criticism of China’s political situation made by the country’s outgoing premier, which highlights growing discontent towards the Xi regime.
The EU wants 40% of their required clean tech to be manufactured domestically.
Short morsels to appear smart at dinner parties
🍱 Cockroach milk, DNA-personalised food, and golden rice: a panorama of the food of the future.
🧫 “Cell painting” is the latest AI tool to accelerate drug discovery.
👶 Using AI to make US presidents look like children has a surprising effect.
🐙 The Scottish government is using its data to design a collective intelligence system (via EV member Claudia Chwalisz).
🦾 Microsoft Germany announced that GPT-4 would be multimodal.
💳 Mastercard is experimenting with using quantum computing.
End note
I spent about 24 hours in airports this week as I travelled to New York and Edinburgh for various appointments. Apologies to NYC-based members for not pulling together one of our amazing meetups. But I was in town for less than 36 hours and packed in ten meetings and some sleep.
As I moved between planes, cabs and meetings, I managed to record a 4-minute reflection on how I view four categories of AI risk.
Cheers,
Azeem
What you’re up to — community updates
Tom Wheeler: “The US and UK establish positions as regulatory first movers while the US watches”.
Martha Lane Fox on a four-day work week, and what it takes to get it right.
Robert Chandler created an iMessage integration using ChatGPT and Whisper models.
Claudia Chwalisz joins the Global Advisory Council of The Data Tank, a new think tank focused on issues around data trustworthiness, reuse, accessibility.
Share your updates with EV readers by telling us what you’re up to here.
I spoke with Ramya Swaminathan, the CEO of Malta, a long-duration energy storage company last year. Listen to our conversation here.
One part missing from the “how did China do it?” Part, and that is a huge multi decade long theft of IP. Even the low estimates of how much value has been taken are astounding.
On ChinAI: https://twitter.com/blader/status/1634089493295931392