Money Thrill : That the  "AI app layer will ultimately determine the value of the industry as the current infrastructure buildout will become a bubble without value creation on the back end". That's a hypothesis ... for example.
The Danish company Nordic, a counterpart of Teleperformance. Nordic is a leading customer contact services company, similar to Teleperformance, offering solutions for customer service, technical support, and other business processes.
The use of artificial intelligence (AI) in customer contact services can deliver significant savings. According to recent reports, the use of AI can increase productivity by 30 to 50 percent
> 1. This is because AI systems, such as chatbots and automated responses, can take over many tasks that would otherwise be performed by employees. This allows companies to significantly reduce their personnel costs. In addition, AI offers the opportunity to improve the customer experience by providing 24/7 support and responding faster to customer questions
>2. This can lead to higher customer satisfaction and loyalty, which can also indirectly lead to cost savings. According to a McKinsey report, AI can automate up to 40 percent of total working hours in the Nordics. By Nordic was it millions of money because it no longer needed 300 employees. When the company announced this, Teleperformance fell to a low point precisely because it was not yet using AI.
So, is that a BUBBLE  ? That's Fundamental business

Money Thrill : I read a book about artificial intelligence and the problem is the PATTERN RECOGNITION in the Black boxes. This requires an enormous amount of human intelligence and commitment and takes months to train. It requires very large memory-intensive capacity. Perhaps $Micron Technology (MU.US)$ is already secretly making this at Imec Belgium, the largest laboratory in the world.  5000 engineers. The nano chips for 2035 exist already? But it is secretly. They have already buyed an Quantum lithography machine for nano chips?
Let me give you an example again. The recognition of clinical pictures with scanner, NMR or radiographic examinations is incredibly promising. Genome research also took 2 days instead of 14 years. One company succeeds, another shall fail. [Photo imec]

Money Thrill : Perhaps Brian,  we must look for the best Foundries and designers? Intel is too late? AMD also? A designer to help is, i suppose $ServiceNow (NOW.US)$? $Micron Technology (MU.US)$ for advanced memory-intensive chips in 2025? Have you other ideas or suggestions ? Nvidia also good and i suppose $Super Micro Computer (SMCI.US)$  also for the Tracks? $Qualcomm (QCOM.US)$ for the Snapdragon? $Arm Holdings (ARM.US)$ for patents or intellectual property rights. ?