January 2, 2026 Editor’s Note: Yogi Shulz is a noted IT professional with decades of a experience. We are pleased to feature his predictions for the coming year. We’d love your comments via the Contact Us form on this site.
Information technology continues to develop rapidly. 2026 will be no exception. Here are my Top 10 predictions. Undoubtedly, as these developments mature, opportunities to apply these technologies more widely across your enterprise will arise.
1. Agentic and autonomous AI systems
One of the most significant IT developments of 2025 was the maturation of agentic AI — autonomous software agents capable of performing complex tasks across industries without explicit human intervention. These systems can make decisions, trigger workflows, and co-manage business processes. Example applications include logistics, customer service, and healthcare.
Agentic AI will profoundly restructure enterprise and government operations with significant changes in processes and related employment.
2. AI bubble? What bubble?
High share prices of AI giants, particularly amid huge losses and no clear path to profitability, continue to attract investors despite much talk of an AI bubble akin to the Internet bubble of 2000.
Huge bond issues to finance billions of dollars invested in AI data centers are weighing on the balance sheets of AI giants, but have not reduced their share prices.
Notwithstanding talk of an AI bubble, giant funding rounds for AI companies, large and small, are fully subscribed. Anthropic, OpenAI and SpaceX are all rumoured to be planning giant IPOs in 2026.
Short sellers, who believe all this AI enthusiasm is “irrational exuberance,” to quote former Federal Reserve Board chairman Alan Greenspan, are being burned.
Explosive growth in demand for AI software-as-a-service (SaaS) suggests sky-high valuations and investments will continue.
3. Nvidia will see credible chip competition
The high cost of Nvidia GPU chips, supply constraints, and the desire to avoid single-vendor dependence are driving the development and adoption of rival solutions. While Nvidia currently maintains a dominant 80%-90% market share in the AI accelerator market, competition will erode this share in the years ahead.
Here’s how major chip manufacturers are responding to compete with Nvidia:
• Advanced Micro Devices (AMD) is selling GPU chips for AI inference tasks.
• Intel is offering chips with performance comparable to Nvidia’s GPUs.
• Qualcomm is selling on-device AI chips for PCs and phones. This approach could slow the growth of cloud inference demands.
• Chinese chip makers are developing their own advanced GPU chips.
Here’s how Big Tech companies are responding to lower their total cost of chip ownership:
• Google has developed Tensor Processing Units (TPUs) as custom AI accelerators.
• Amazon (AWS) has developed its own Trainium chips for training AI models and Inferentia chips for inference.
• Meta (Facebook) has designed its own custom chips, the Meta Training and Inference Accelerator (MTIA) chips.
• Chinese companies are accepting the additional operating costs of older chip generations until newer Chinese chip generations become available to them.
Given the incredible forecasts for AI processing growth, all players will sell many chips as they also continue to improve the price/performance of their GPUs.
4. Apple will improve its AI implementation
Apple Intelligence is Apple’s personalized AI system, deeply integrated into its operating systems. Across multiple releases, Apple has improved Siri, Writing Tools, and Image Playground while maintaining its focus on privacy. However, Apple Intelligence remains far from replacing comparable offerings from other AI giants and app developers.
Apple will continue developing Apple Intelligence relentlessly as it competes with Google’s Android devices and other app developers.
5. Denying advanced chips to Chinese companies
Efforts to keep advanced GPU chips out of Chinese companies’ hands will be more of an irritant to relations with the US than a hindrance to Chinese progress in the AI race.
Chinese companies will sufficiently circumvent the restrictions and compete successfully in the AI race with the following actions:
• Dramatically reducing computing resource consumption for AI model training, as DeepSeek demonstrated last year.
• Designing more advanced GPU chips in China, as HiSilicon (Huawei’s chip arm), UNISOC, Alibaba, and Baidu are doing.
• Strengthening the Chinese software ecosystem and related AI frameworks. Government incentives are accelerating this work.
US regulatory interference will not help US-based AI companies maintain their lead in the AI race. Only continued innovation will do that.
6. SLMs will finally receive attention
Rarely discussed Small Language Models (SLMs) will finally receive more attention. SLMs are efficient, domain-specific AI models that offer these advantages:
• SLMs deliver high accuracy with no bias or hallucinations for niche AI applications.
• The small number of required parameters in SLMs makes it feasible to curate AI training data thoroughly.
• SLMs are much easier and cheaper to train than LLMs, thanks to their fewer parameters.
• SLMs offer fast, low-cost inference performance due to their small number of parameters.
• SLMs run on small, cheap computing environments, sometimes even without high-end GPUs.
Companies will deploy SLMs when LLM-based AI applications disappoint.
7. Electricity access will hamper AI data centers
Electricity shortages in the US and the glacial pace of grid connection approvals will hamper the startup of some AI data centers.
The heavily regulated electricity industry will delay the deployment of AI data centers that rely on the grid. There are multiple issues:
• The electricity industry is not designed to respond to the rapid growth in demand that AI data centers represent.
• Consumer are panicking that AI data centers may increase their electricity costs. They will lobby regulators, who will slow approvals.
• Permitting fights about rights-of-way for new high-voltage transmission lines.
• Permitting fights about building new electricity generation facilities, especially if they propose to emit any GHG emissions.
• Community concerns that AI data centers generate excess noise and consume too much water.
For AI data center developers, these issues create delays and added costs. As a result, developers will build more AI data centers in rural areas with on-site electricity generation.
8. No business case for AI applications
Companies are struggling to build compelling business cases for various AI applications. The following issues are the typical impediments:
• Poor data quality undermines AI model output and is expensive to correct.
• The high cost of AI application projects.
• The lack of robustness of new AI SaaS) vendor solutions.
• The lack of available talent with sufficient experience in AI applications.
Companies will overcome these issues and identify killer AI apps.
9. Profitability of AI SaaS solution providers
The main factors that are influencing the profitability of AI SaaS solution providers include:
• Improving AI training and inference efficiency vs. increasing AI model size.
• Decreasing unit cost of computing resources vs. increasing consumption of computing resources.
• New customers contributing to economies of scale vs. increased competition among AI SaaS solution providers, leading to lower prices.
• Improving AI-driven software development vs. the cost of software development teams.
The price-performance of AI models will continue to improve, undermining the path to profitability for both AI giants and their smaller cousins.
Several serious efforts to build AI data centers in space will proceed despite considerable skepticism and even derision. The attractions of space include:
• Unlimited cheap solar power.• Superior and cheap cooling via space’s vacuum.
• Absence of terrestrial security threats.
• Vast scalability.
• Reduced environmental impact.
The high cost of transporting hardware components into space, even after the significant cost reductions SpaceX and its competitors are offering, and the difficulties of making service calls, will not deter proponents.
These information technology developments and many more will make 2026 an exciting year that opens up new application possibilities. Stay tuned.
