🔮 Solar saves us; AI risk; Nvidia, the chip king; cyberattacks and kissing++ #424
Your insider's guide to AI and exponential technologies
Hi, I’m Azeem Azhar. As a global expert on exponential technologies, I advise governments, some of the world’s largest firms, and investors on 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:
Solar and batteries lead the post-fossil transition;
Nvidia became a trillion-dollar company.
Sunday chart: The sun is rising on net zero
Solar deployments and battery manufacturing are growing much faster than expected, and it’s looking like paired together, they could help us achieve net zero carbon emissions by 2050.
The IEA’s latest assessment reveals that the growth of clean energy production is escalating so rapidly that we’re finally on track to meet the 2030 objectives to get to the 2050 net zero goal. Solar and battery are the main drivers.
In fact, solar capacity is set to easily surpass the deployment demands projected by the IEA’s model for 2030. Meanwhile, battery manufacturing potential has leapt from 6% to 97% of net zero levels (including Q1 announcements), and electrolyser capacity has jumped from 4% to almost 60%.
The IEA report indicates that current solar factories are operating at a 40% utilisation rate, implying that they are collectively generating less than half of the total photovoltaic (PV) modules that they could potentially produce. If all existing and upcoming factories maintained this rate, they would generate approximately 650 gigawatts of modules annually, which is in line with the IEA’s net zero pathway for 2030.
The clean energy investment increase is being led by China, followed by the European Union and the United States.
Simon Evans’ tweet thread on the report is worth reading.
Key reads
Risky business. Major AI companies are coming out with recommendations on how to manage AI risks, the last of which was Google Deepmind’s paper arguing that model evaluation is critical to detect “extreme risks”. The idea is that if we can determine whether a model is capable of doing harm (dangerous capabilities) and is likely to actually do the harm (alignment), we could take appropriate action. One important criticism of this approach is that it imbues models, rather than the people, or corporations, who run them with agency. Jeni Tennison makes that case well and also calls for a “deanthropomorphisation” of these tools.
I had earlier recommended democratic engagement through citizens’ juries or deliberative mechanisms to discern values and preferences. OpenAI seems to agree. The company announced that it would fund “experiments in setting up a democratic process for deciding what rules AI systems should follow”.
I’m glad we agree. However, companies should not be defining the rules of the brave new AI world, while governments play catch up. Second, tech firms haven’t shown themselves committed to tackling problems with AI systems. Consider how the biggest firms haven’t yet fixed racist bias in software first identified years ago, choosing, instead, to hide certain appellations rather than solve the issues. Transparency will be crucial in ensuring AI governance is profoundly just and sensible.
Chip rush. The AI revolution has made demand for microchips explode, and chip manufacturing companies are more valuable than ever. After announcing its Q1 financial results, Nvidia has become only the fifth publicly-traded American company to be worth $1 trillion, a feat only managed by the likes of Apple, Microsoft, Alphabet, and Amazon. State-of-the-art generative AI models need thousands of GPUs for their operation - and Nvidia controls about 88% of the GPU market. Nvidia added more than $150bn in market cap in one day. Intel, the chip giant of the PC era, only has a market cap of $120bn. The firm blew mobile and appears to have missed the transition to AI too! (For a longer read on Nvidia’s journey, I recommend this analysis.)
Weekly commentary: What AI bosses need to hear
This week I had a good discussion with Sam Altman, the CEO of OpenAI, as a part of his global listening tour. I can’t remember ever coming across an industry sector where leaders were prima facie as enthusiastic about introducing regulation to their sector as AI executives do today. That, I think, is a positive development. However, the question inevitably arises: how will these leaders react when the rubber hits the road?
🐙 What AI bosses need to hear
Watch now (5 min) | In the past week, I engaged in a compelling discussion with Sam Altman, the CEO of OpenAI, as a part of his global listening tour. Our conversation will be released as the first episode of my new TV series Exponentially, which will be coming to your screens in July.
Market data
According to a survey, six in ten adult Americans have heard of ChatGPT, but only 14% of them have tried it.
The Tesla Model Y was the best-selling car in the first quarter of 2023, a first of an electric car.
Swedish internet provider Bahnhof plans to use small nuclear reactors to power its data centres and sell surplus energy to the city grid, powering 30,000 households.
The components of Apple’s forthcoming extended reality headset will cost around $1400 according to a leaked Bill of Materials.
Clean energy investment continues to grow faster than fossil fuel investment.
Russia’s plans to diversify gas supply with China’s help are not going well.
Short morsels to appear smart at dinner parties
🧠 A new brain-spine interface successfully restored movement in a paralysed individual.
🔬 Using AI, scientists have identified a promising new antibiotic, abaucin, from thousands of potential chemicals.
✈️ France has banned short-haul flights, provided the same trip can reasonably be made by train.
💋 The recorded history of kissing is extended by 1,000 years with evidence found in ancient Mesopotamia from at least 2500 BC.
🔎 “Explaining the how”: BBC verify is the network’s new transparent, fact-checking faction.
🏴☠️ Beware of the prompt injection cyberattacks.
🎭 A crypto governance scheme was hijacked by using fake Twitter accounts for fraudulent voting.
End note
Hidden in my Weekly Commentary is a short announcement about the new TV series, Exponentially with Azeem Azhar, that I am hosting.
The show kicks off in July as part of the Bloomberg Originals slate. It’ll be available on Apple TV, Amazon TV, Samsung TV and other connected devices, as well as on linear TV (on Bloomberg and BTV+) and Bloomberg.com
We’ll be sending out more updates—but I do hope you’ll all keep your eyes peeled for it.
A
What you’re up to — community updates
Meredith Whittaker writes a challenging critical review which connects digital computing, industrial labour and slavery—and their relationship to the work of Charles Babbage.
Matthias Ljungman’s VC Moonfire Ventures raises a $90m fund for early-stage startups and a $25m opportunities fund.
Laure Claire and Benoit Reillier are running the Platform Leaders event on June 7th on generative AI, platform regulation and the circular economy.
Ramsay Brown and his team at Mission Control announced a visual no-code builder for trustworthy generative AI.
Robbie Stamp talks participatory democracy and friendship with Douglas Adams in a Q&A with DemocracyNext.
Share your updates with EV readers by telling us what you’re up to here.
To be fair, since they only worked out how to curtail LLM's excesses recently and nobody truly understands how they work. The only way to get rid of bias is to remove it from the source material, and I'll wager that nobody wants a single company to be deciding what goes in and what doesn't.