Cloud Tracker Pro

CTP Q2 Update: AI Infrastructure Shifts Toward Inferencing

At the halftime mark, 2024 continues to be about artificial intelligence (AI), as hyperscalers escalate spending on infrastructure and enterprises ponder next steps. These are some of the highlights of our latest Cloud Tracker Pro premium report: AI Flows Toward the Mainstream.

As interest in AI takes hold in enterprises, they are likely to consider different architectures and methods of implementing AI infrastructure. Futuriom believes the market may now shift towards edge infrastructure and inference, a market Futuriom expects to nearly double within the next two years. Inference refers to adapting large language models (LLMs) to enterprise applications through exposure to company- or industry-specific datasets.

Inference is anticipated to be a large and lucrative market, though it will take time to fully materialize. Challenges include the quality and maturity of data used, the design of AI-enabled applications, and the infrastructure of networking, compute, storage, and security used in implementations.

CTP subscribers can download the report here.

Highlights of This Report

This Cloud Tracker Pro (CTP) report, available exclusively to CTP subscribers, highlights the trends we saw in the first half of the year in the building of cloud infrastructure. Here are the highlights that we cover:

AI interest is high among enterprise management. But strategies for managing its adoption remain in flux.
Data is essential to the growth of inferencing. But the process of adapting AI models to fit enterprise applications depends on mature, well-managed and secure data stores.
A market for AI edge equipment is growing as demand for inferencing requires performance and security. Futuriom expects this market to nearly double within the next two years.
Ethernet is likely to advance as an alternative to InfiniBand. Demand for Ethernet in AI networks speaks to the wider needs of companies for networking that is cheaper, doesn’t depend on one vendor, nor requires specialized training.
The future of sovereign AI is questionable. Government data is prone to political influence and conflict, and in authoritarian regimes its use in AI could represent a threat to citizens’ rights.
Private AI represents the future of enterprise AI architecture.