Investing in and Using AI
Nvidia is currently worth over $2 trillion and is leading frenzied investor interest in AI. Yet this is only Act 1 of the AI story. What do Acts 2 and 3 have in store?

Act 1:  The Enabling Technology
The idea of AI has been around for decades. But it had to wait for computing power and database storage to become a practical tool for businesses and consumers alike.

Nvidia made a crucial advance combining computing and application platforms in CUDA (Compute Unified Device Architecture). This enables much faster general processing and results in the company selling over 70% of AI chips.

Nvidia’s GPUs (Graphics Processing Units) perform simultaneous calculations while traditional units do them one at a time. The waiting list for its quarter of a million-dollar Hopper GPU is months long. This is both an opportunity and a risk.

Whenever demand for a product is this strong then competition is fierce. Other companies are catching up and the largest customers want their own chips. Sanctions on China mean the country’s leadership is focused on building its own version of cutting-edge technology. And Nvidia’s units are built in Taiwan which carries medium-term geopolitical risks.

The forgotten child in this story is data storage. Vast new warehouses using new technology are required to support the development of AI. This creates opportunities to invest in land, power, and the technology itself.

Data warehouse operators must be experts in data tiering and thin provisioning. Tiering is the prioritisation of data based on demand. Provisioning uses capacity as required, enabling systems to handle more volume than their physical limits.

When mobile telephones exploded onto the scene, chip manufacturers such as Qualcomm shot to prominence. Today the company is less than 10% the size of Nvidia. Qualcomm starred in Act 1 of mobile telephony and bigger players were about to enter the scene.

Act 2: The Core Infrastructure
The second layer of AI technology is the models built on Nvidia’s chips. The most prominent is OpenAI’s ChatGPT, which is part owned by Microsoft. But Google’s Gemini and Anthropic’s Claude AI are significant competitors more skilled at certain tasks.

These are large language models that comprehend words from text. There are models for computer vision and sound recognition that may become more important. That still leaves the more complicated challenge of creating taste, touch, and smell to match human senses.

Phone manufacturers such as Apple and Samsung overturned the mobile industry with smartphones. Blackberry and Nokia were turned to dust. The evolution of AI models is in its infancy and the eventual winners are unknown.

Given the importance of model performance, only a handful of companies may triumph in this space. But almost every company has a role to play in Act 3.

Act 3: The Application Layer
Users could not do much on the first smartphones. Apple built a few apps and released the technology to cool kids and those who revel in early adoption. The mass market uptake of iPhones came when the system was opened up to outside app developers.

Social media and search companies such as Meta and Google were the biggest beneficiaries of application development. But most companies have a web page, and many have an app for clients.

There is a view that the costs and complexities of AI mean the winners in Act 3 will be existing large companies. But at RemoraTech we are working on another theory.

Many countries want to develop their own AI models to reduce reliance on potential rivals. Several alternatives are open source, which increases their cost effectiveness for users. This opens the world of AI to small and medium enterprises (SME).

The primary purpose of AI for these firms is to serve customers better. This means personalising experiences and improving service in a way that is impossible without AI. All it requires is fine tuning of open-source models using in-house data and documentation.

Act 3 of the AI story will see smart businesses flourish. This includes chatbots for the insurance industry with pinpoint recall of customers’ policies. It means tailored reports for asset management clients created as often as required. And it enables sourcing better deals faster for venture capital and private equity, using precise investment criteria.

A Starring Role
At RemoraTech we’re focused on enabling Act 3 for SMEs. This allows them to compete on a level playing field with much larger competitors. Let us know if you dream of starring in AI’s final chapter.