Q* + Tigris = ?
Q* + Tigris = Altman's AGI Superintelligence Integrated Supply Chain
Investors are missing the point of the Sam Altman OpenAI drama. It wasn’t about who gets to lead the company. It was about a grand vision of building an integrated supply chain that supports AI, regenerative AI, and even an AGI Superintelligence. Reports are surfacing that Altman announced that Open AI was on the brink of achieving a significant breakthrough only the day before his firing. Apparently, some of the staff at Open AI wanted the Board to know that Altman’s breakthrough, though an algorithm known as Q* (pronounced Q-Star), could “threaten humanity.” But, for Altman, the breakthrough was not enough. He also wants to build a whole integrated supply chain to empower his new superintelligence. The Open AI leadership drama is about his vision of marrying a superintelligence AI with an integrated supply chain of chips, phones, robotics, and the world's largest collections of data and LLMs (large language models). Its working name is Tigris.
Let’s add the up pieces of Altman’s vision:
Altman envisages AI chips and phones designed by Sir Jony Ive's firm LoveFrom (whose website is beautiful) See: Jony Ive, Sam Altman wants to create the ‘iPhone for AI’ and “OpenAI and Jony Ive in talks to raise $1bn from SoftBank for AI device venture: ChatGPT creator in early discussions to create the ‘iPhone of artificial intelligence’. Rumours are flying that Maysoshi Son is ready to back their collaboration with Cerebras Systems, who make the fastest AI chips in the world, should make something specifically for Open AI. The firm won the coveted Gordon Bell prize in 2022 for making the computer hardware that permitted fast analysis of the SARS-CoV-2 virus. As the NYTs reported, They built the first wafer-scale chip and announced a “mind-boggling” 7 nano meter chip with “2.6 trillion transistors and 850,000 AI optimized cores. The largest graphics processor unit (GPU) has only 54 billion transistors – 2.55 trillion fewer transistors than the WSE-2. The WSE-2 also has 123x more cores and 1,000x more high performance on-chip high memory than GPU competitors”, according to Cerebras. Wired wrote in 2021 that Cerebras’s swarm X software allows the chips to form a “cluster” of chips that “can run AI models that are more than a hundred times bigger than the most gargantuan ones around today.” That cluster capability is much bigger and faster today. These are the chips that are needed for the metaverse, neural networks and “massive parallel mathematical operations.” “The neural network behind GPT-3 has around 160 billion parameters.” “GPT-4 will be about 100 trillion parameters.”
All this means we are witnessing the end of chip miniaturization and the start of a new generation of chips that are designed to handle the overwhelming processing demands of massive data sets, virtual reality and AI. It means Cerebras hardware will now compete with NVIDIA. Cerebras is building a supercomputer for G42. Now enter Softbank. Soft Bank owns 90% of ARM, The UK chip maker whose recent IPO enriched Soft Bank’s coffers and helped Softbank overcome the bad press around WeWork. The FT recently said this of Mayasoshi Son, the head of SoftBank: “Son has described Altman as “one of the key people on Earth” and said he speaks to him almost every day.” Son is part of “A Silicon Valley Supergroup Is Coming Together to Create an A.I. Device,” according to the NYTs. That group includes Open AI, LoveFrom, Cerebras and G42.
G42 is an Abu Dhabi-based AI firm (backed by Mubadala and Silver Lake) that has a range of AI firms it fosters. Altman announced that Open AI would be collaborating