Exponential View

"A zero-carbon electrified economy is both cheaper and better than a fossil-based economy. An electrified economy is bad news for autocrats and traditional power structures." Clinging on to this thought from the excellent Exponential View

Exciting news! ?? I'm thrilled to share the joint research on the future of computing power I've conducted with Azeem Azhar and Nathan Warren from Exponential View, and Riccarda Joas and David Zuluaga Martínez from the BCG Henderson Institute. Many people worry that GenAI adoption will result in a computing power crunch. We ran the numbers and here’s what we found: - We're more likely heading towards compute abundance, not scarcity - By 2028, GenAI workloads will likely use only ~34% of the global AI computing supply - Energy and not hardware is more likely to constrain computing power supply Our analysis (read it here: https://lnkd.in/eVVdmSaH) shows that it is crucial for business leaders to understand that they need to plan for a future of orders-of-magnitude more compute, a future wherein AI will be able to conquer increasingly complex tasks. Stay tuned for Part 2 next week, where we'll explore strategies to capitalize on this compute abundance! What's your take on compute abundance? How might it reshape your industry? #AICompute #FutureOfTech #StrategicForesight #GenerativeAI #GenAI #ArtificalIntelligence

The world's computers consume vast amounts of energy, and this consumption is expected to triple by 2030. This comes on top of existing pressures on the grid. We know that energy efficiency gains in compute are reaching an impasse within the current computing paradigm, but what is the way forward? One potential solution can be found in reversible computing. I'll try to explain it simply: computers we use today perform operations that are irreversible, meaning once they compute something, you can't easily go back to the previous state. This irreversibility is what causes heat and energy loss. In contrast, reversible computing allows operations to be undone. Think of it like having a movie where you can watch scenes both forward and backward without losing any detail. By making operations reversible, computers can significantly reduce the amount of energy wasted as heat. This is because reversible operations can, in theory, recover and reuse energy that would otherwise be lost. The idea ties back to basic physics principles. The second law of thermodynamics says that systems naturally become more disordered over time (entropy increases), which typically results in energy loss as heat. Reversible computing tries to minimise this disorder and energy loss. With reversible computing, computers would generate much less heat. This is important because too much heat can interfere with computer performance. And as computers become more efficient, the demand for electricity to power them would decrease, which is crucial given the growing global energy needs. So, how could this all work out? David Galbraith wrote an essay explaining the science and history behind reversible computing in great detail for Exponential View. It is a must-read if you want to understand what alternative computing paradigms could look like in the near future. I'll link David's piece in comments. Image via DALL-E

"Breaking the energy barrier with reversible computing" an Exponential View guest post by David Galbraith ?? https://lnkd.in/emuE6EPE

I spoke with Azeem Azhar of Exponential View about: - How he thinks about the speed of progress in AI - Inevitability narratives, cultural affordances, centralization/decentralization - Future outlooks

Our 30 days of Charter 30 continues with Azeem Azhar, founder of Exponential View. Azhar's ability to distill complex topics into understandable concepts makes him a must-follow for anyone interested in technology and the future of work. His Substack, Exponential View, helps readers make sense of the latest updates in AI through great analyses and interviews with top-tier researchers and technology leaders. Read more about Azhar, and view our full list of Charter 30 honorees: