Orkestro "The Biggest Barrier to AI Adoption is GPU Cost"

[Financial News] Artificial Intelligence (AI) and cloud software specialist company Orkestro announced on the 17th that it released the results of a survey on 'AI Utilization in Cloud Environments' conducted last June with 6,615 employees from companies and public institutions using cloud services.
The biggest constraint when adopting AI was △ the cost of technology adoption such as Graphics Processing Units (GPU) (23.5%). This was followed by △ lack of professional manpower (22.6%) △ data security concerns (14.4%).
In terms of expectations for AI, 'cost optimization' and 'real-time security response' were the highest at 20.7% each. 'Failure cause analysis' (17.1%) and 'performance bottleneck resolution' (15.9%) followed. Responding companies cited △ cost optimization (22.7%) △ security management (19.1%) △ failure response (15.2%) as major tasks. The tasks that required the most time were also △ cost analysis and optimization (22.7%) △ monitoring and failure response (19.5%) in order.
Operational challenges showed distinct differences depending on the size of the company. Large companies focused on security management and cost analysis, medium-sized companies on infrastructure configuration and monitoring, and small companies on cost optimization.
There was also a large gap in the capacity to invest in Graphics Processing Units (GPU). Medium-sized companies were concentrated in the range of 5 million to 20 million won per month (33.3%), while large companies focused on more than 100 million won per month (34.5%).
The cloud infrastructure structure is becoming increasingly complex. In this survey, 49.1% of all responding companies said they are using multi or hybrid clouds. Just a few years ago, most companies relied on a single cloud, but now nearly half of the companies are utilizing multiple clouds. Among them, 60.2% were found to be operating three or more clouds simultaneously.
Orkestro is reducing operational burdens through its integrated management solution 'Orkestro CMP', which encompasses public, private, multi, and hybrid clouds. It is also promoting the 'Cloud for AI' strategy to lower GPU infrastructure costs and AI adoption barriers.
Orkestro proposed the server virtualization solution 'Contrabass' as a solution to the GPU infrastructure problem, which companies feel is the biggest burden in AI adoption. In AI environments that require high-performance computing, efficiently and stably utilizing GPU resources is key.
Contrabass provides GPU virtualization, optimization, scheduling, and monitoring functions to ensure excellent operational efficiency and stability even in complex cloud environments, Orkestro explained. Companies using VMware (virtualization software) environments can perform migrations without system downtime through 'Contrabass Legato', reducing the burden of replacing existing systems.
In addition, it provides an AI environment encompassing everything from infrastructure operation automation to information search through the generative AI chatbot 'Clarinet' and the knowledge search platform 'Oboe (G-AIDSP)'. Clarinet is a query-based conversational solution that automates repetitive tasks, and G-AIDSP reduces the chronic hallucination problem of generative AI to enhance reliability.

mkchang@fnnews.com Min-kwon Jang Reporter