AI's Single Point of Failure – Rob Toews – Ted [video](youtube.com)
youtube.com
AI's Single Point of Failure – Rob Toews – Ted [video]
https://www.youtube.com/watch?v=AJGrdtKT3LM
2 comments
"The world's most important advanced technology is nearly all produced in a single facility," says AI expert Rob Toews.
While it is not a given that the future of AI will be dependent on quantum computers (e.g. Classical AI is likely to continue its progress, quantum computers might be best suited for specific tasks rather than a complete replacement for classical AI), every indication is that quantum computing will address the current limitations in AI with quantum algorithms tackling specific problems more efficiently than classical algorithms, potentially leading to faster training of AI models, better decision-making, and more efficient resource utilization.
Assuming a dominant role of quantum computing in the future of AI, it is interesting to note that quantum computing is not as dependent on the 3 nm chip foundries of TSMC. Let's look at the leading quantum computing developers:
1) D-Wave's quantum annealing processors rely heavily on custom-designed CMOS chips manufactured by TSMC at the 7nm and 16nm nodes.
2) IBM's quantum processors, like the IBM Quantum System One, utilize a mix of commercially available and custom-designed chips. While some components might come from TSMC, others could be sourced from other foundries like Samsung or Intel. IBM has also invested in developing its own chip fabrication capabilities, reducing reliance on external foundries in the long run.
3) Microsoft's quantum processors primarily utilize commercially available chips, though they are exploring custom chip designs for future generations. TSMC's 3nm capabilities could offer advantages in terms of density and power efficiency, but it's not a critical dependency for Microsoft's current systems.
4) Google's Sycamore quantum processor, which achieved quantum supremacy, relied on custom-designed chips fabricated at Google's own facilities. They continue to invest in their internal chipmaking capabilities, reducing dependence on external foundries like TSMC.
5) Rigetti's focus on three-qubit superconducting processors minimizes the need for complex chip fabrication. While they might utilize commercially available components, the overall dependence on TSMC's 3nm foundry is minimal.
While TSMC's 3nm foundry capabilities offer potential benefits for some quantum computer developers, it's not a universal dependency. Developers are diversifying their chip sources and exploring alternative technologies to reduce reliance on specific foundries. Developers using alternative qubit technologies like trapped ions or topological qubits might have even less reliance on traditional chip foundries like TSMC. At the same time, as quantum hardware matures and scales, the need for custom, cutting-edge chips might increase, potentially leading to greater reliance on advanced foundries like TSMC.