Dividends and Corporate America turns to open-source AI

We all know AI is here and every year, more AI and AI is being used, but what type of AI is being used?

In an article by Eli Tan of the New York Times News Service, AT&T is a typical company, it is using AI to help with customer service, call transcriptions and coding and it turned to AI models of Anthropic and Open AI. The AI models charged fees and were “closed”, which meant the AI companies did not share their underlying code.

Then AT&T changed. As AI costs have gone up, Andy Markus, the company’s chief data and AI officer, looked for less expensive alternatives. Mr. Markus moved to ‘open” models.

By May, open models accounted for 20% of AT&T’s AI use. That has risen to 40% and then jumped to 60%. AT&T’s was saving up to 80% on AI costs compared with a year earlier.

Open-source AI has become all the rage in corporate America. Companies such as AT&T, Airbnb and Deloitte are turning to open AI models that are easy to customize and cheaper to use. Last month, open models accounted for 58%, up for 10% a year earlier, according to US user data from Open Router.

Nvidia announced it was buying Hugging Face, which owns a library of open AI models, for $12.9 billion.

Open models come in 2 types – one is an open-source model which makes its underlying code public. The other is open-weights model, which makes its weights – the numeric calculations that determine how the model reasons public.

Jerry Tang, CEO of Atlas Cloud estimated that Chinese open models are 80 to 90% powerful as OpenAI and Anthropic. He used a sporting car analogy – Mazda does a fine job getting you there, but Maseratis are simply not worth the investment.

The good thing about AI means that IPOs for OpenAI and Anthropic maybe in trouble, because they need to charge more money for research and development.

Many open models can now be powered by a phone or laptop, according to David Stout, CEO of WebAI.

Closed models are best for heavy-duty tasks such as coding, image and video generation, while open models often excel at simple, specialized tasks.

Back to AT&T, the telecom company takes open-weights models and customizing them to make new tools that can transcribe calls and facilitate customer service.

AT&T researchers study Chinese models, but they are not using them. At the moment, AT&T is using Gemma AI model from Google and Llama AI from Meta.

Linking to dividend paying stocks, often people discover them through cycles, many people buy a company hoping to be rich today, but real wealth typically takes some time. Eventually, people look for quality and with quality comes dividends and total returns. Given the rule of investing is try not to lose money, it is important to use your time to evaluate and continuous evaluate (or do your homework of the alternatives), then when you decided to invest more money you will have many ideas of what to do, given all the options available to you.

There are more questions than answers, till the next time – to raising questions.

Leave a Reply