As generative AI adoption accelerates, many companies are struggling with the handling of confidential information. Many executives hesitate to input customer data or internal sensitive information into cloud-based AI services.
Amid this, SB C&S has begun offering an on-premise generative AI platform. This move signifies more than just a product launch. For executives, a “second option” for AI utilization has become a reality.
The Risks of Cloud AI and the Value of On-Premise
Cloud-based generative AI like ChatGPT and Claude is certainly convenient. However, the possibility that input data could be used for training, or the risk of information leaks, cannot be completely eliminated. Especially when handling customer information, unpublished business plans, or patent-related data, sending it to the cloud can itself pose compliance issues.
The on-premise platform that SB C&S has started offering solves these challenges. Since the AI model runs on the company’s own servers, data never leaves the premises. For executives looking to apply generative AI to highly confidential tasks, this is a significant step forward.
I personally use AI for automating contract reviews and background checks, but initially, I was worried about whether it was safe to send this data to the cloud. With an on-premise environment, you can enjoy the benefits of AI without those concerns.
Accurately Understanding Costs and Implementation Hurdles
When considering on-premise AI, cost is a key concern. SB C&S’s service requires an initial investment for purchasing servers and GPUs. Depending on the model, this often involves an investment of roughly $7,000 to $35,000.
However, running costs may be lower than cloud-based options. There are no API usage fees or pay-as-you-go charges based on data transfer volume. In the long run, if you’re paying tens of thousands to hundreds of thousands of yen (hundreds to thousands of USD) monthly for cloud services, the investment can be recouped in one to two years.
When implementing, the availability of in-house IT personnel capable of managing GPU servers is also a crucial factor. While sales partners like SB C&S often provide implementation support, daily operations will need to be handled internally.
Choosing Between On-Premise and Cloud Is a Strategic Turning Point
The key point here is that it’s not a binary choice between “on-premise or cloud.” In practice, it’s more realistic to use a combination based on the nature of the task.
For example, use cloud AI for low-sensitivity tasks like creating external marketing copy or conducting market analysis based on public information. Meanwhile, adopt on-premise AI for tasks involving customer data analysis or internal financial information. This hybrid strategy optimizes the balance between risk and cost.
According to a PwC survey, Japanese companies are lagging behind other countries in “generating results” with generative AI. One reason is hesitation to use AI at all due to security concerns. The emergence of on-premise solutions has the potential to break down this barrier.
Concrete Implementation Steps
As an executive, the first step should be “mapping your company’s data.” Identify which tasks handle what kind of data, and which of those you want to process with AI. Then, consider implementing on-premise AI for tasks involving highly sensitive data.
With SB C&S’s service, you can choose open-source large language models like Llama 2 or Falcon. These are available for commercial use and offer a high degree of customization freedom. In theory, you can also fine-tune the model to suit your company’s specific needs.
However, fine-tuning models requires specialized knowledge and GPU resources. I recommend first using the existing model as-is to verify its applicability to your tasks. In my experience, general-purpose models often provide sufficient accuracy initially.
Actions Executives Should Take Now
SB C&S’s announcement shows that the on-premise AI option is now “open to general companies.” The era has arrived where a secure AI environment, previously only feasible for large corporations, is now available to small and medium-sized enterprises at a realistic cost.
First, request a demo of on-premise AI from SB C&S or other vendors. By seeing the actual operation and implementation costs, you can make a decision that suits your company. At the same time, verify how much of your tasks can be covered by cloud AI to smoothly design your hybrid strategy.
AI adoption is no longer in the “whether or not to do it” phase, but in the “how to do it” phase. Now that the on-premise option is available, it’s time for executives themselves to lead a strategy that maximizes the benefits of AI while protecting their company’s data.
Utilizing AI without exposing confidential data—this new option could be the catalyst that propels Japanese companies’ AI adoption to the next stage.
Comments