Summary
HP is changing how businesses use artificial intelligence by focusing on local hardware instead of relying only on the cloud. Jerome Gabryszewski, a top manager at HP, explains that many companies struggle to turn their data into useful AI tools because of messy internal systems and high costs. By using powerful on-site computers, businesses can keep their private information safe and reduce the massive bills associated with cloud services. This shift helps companies move from just testing AI to using it every day in a safe and affordable way.
Main Impact
The biggest impact of this shift is the move toward "local AI" compute. For a long time, businesses thought they had to send all their data to the cloud to run smart models. However, HP is showing that high-performance workstations can handle these tasks right in the office. This change reduces the risk of data leaks and gives companies more control over their budgets. Instead of paying for every single AI request in the cloud, businesses can buy their own hardware and run models as much as they want for a fixed cost.
Key Details
What Happened
Ahead of a major technology event in San Jose, HP shared its strategy for helping large companies manage AI. The company pointed out that while data is often called "the new oil," it is very hard to use if it is scattered across different departments. HP is introducing a range of new hardware, such as the ZGX Nano and the ZGX Fury, which are designed to run massive AI models without needing an internet connection. These machines allow teams to build and test AI tools faster and more securely than using shared cloud platforms.
Important Numbers and Facts
- Spending: Businesses spent $37 billion on generative AI in 2025.
- Budget Issues: About 80% of companies spent more on AI than they planned, often missing their targets by 25% or more.
- Cost Savings: Using on-site hardware can be 18 times cheaper than the cloud over a five-year period.
- Hardware Power: The ZGX Nano is a tiny computer that can handle AI models with 200 billion parameters.
- Future Growth: By the end of 2026, experts predict that 40% of business applications will use AI agents to perform tasks.
Background and Context
In the past, most AI work happened in the cloud because the models were too big for normal office computers. However, cloud costs have become a major problem for many businesses. Every time an AI answers a question, it costs money. Additionally, many companies are afraid to upload their secret business plans or customer data to cloud-based AI providers. They worry that their private information might be used to train other models or could be stolen in a data breach.
HP is solving this by bringing the "intelligence" to the data. Instead of moving data to the AI, they are moving the AI hardware to where the data lives. This approach uses a method called Retrieval-Augmented Generation, or RAG. This allows an AI to look at a company's private files to find answers without ever sending those files outside the building. It keeps the information private while still giving employees the benefits of a smart assistant.
Public or Industry Reaction
The tech industry is starting to realize that the "cloud-only" approach is not sustainable for everyone. Many IT leaders are looking for ways to balance their spending. While the cloud is still great for starting new projects quickly, the cost of running those projects every day is forcing companies to look at on-site options. Industry experts note that the role of IT teams is also changing. Instead of just fixing broken computers, IT staff are now becoming "governors" who decide which AI tools can be trusted with company decisions.
What This Means Going Forward
In the next few years, we will likely see a "three-tier" model for business technology. Companies will use the cloud for very large tasks that happen only once in a while. They will use their own on-site hardware for daily AI work to save money. Finally, they will use "edge" compute for tasks that need to happen instantly, like on a factory floor or in a retail store. This hybrid approach will help businesses stay competitive without going broke.
There is also a growing focus on AI governance. This means setting rules for how AI behaves. As AI models start to update themselves automatically, companies must watch out for "concept drift," which is when an AI starts giving wrong answers over time because the data has changed. Businesses will need to treat AI updates with the same caution they use for any other important software change.
Final Take
The future of business AI is not just about having the smartest software; it is about where that software runs. By moving AI work back to local hardware, companies can regain control over their data and their spending. HP’s focus on powerful, compact workstations shows that the era of relying entirely on the cloud for innovation is ending. For a business to succeed with AI, it must build a strong foundation of its own hardware and clear data rules.
Frequently Asked Questions
Why is local AI hardware better than the cloud for some businesses?
Local hardware is often cheaper in the long run because you pay for the machine once instead of paying for every AI request. It also keeps sensitive company data inside the building, which is much safer than sending it to an outside provider.
What is "concept drift" in AI?
Concept drift happens when an AI model becomes less accurate over time. This usually occurs because the real-world data it is looking at has changed, but the model is still using old patterns to make decisions.
How does HP's RAG system protect private data?
Retrieval-Augmented Generation (RAG) allows an AI to read a company's internal documents to find specific answers. Because this happens on local hardware, the documents never leave the company's secure network, and the AI does not "learn" the secrets in a way that others could see.
