The rapid advancement of AI is reshaping industries at an unprecedented pace. AI-driven solutions are being deployed for data analysis, decision support, predictive modeling, and other high-impact use cases. As enterprises move forward, a critical question remains: What should they anticipate in the next phase of AI evolution?
Emerging AI Models - Small Language Models (SLMs) and Multimodal Models
While some companies pursue the development of large, general-purpose models, others are turning to more specialized approaches, such as Small Language Models (SLMs) and multimodal models, which promise greater efficiency and adaptability.
Small Language Models (SLMs)
SLMs are optimized for specific tasks, which offer focused performance at a lower cost. Trained on domain-specific data, these models are helping businesses solve particular challenges more efficiently. For instance, SLMs can provide actionable insights for inventory management that would otherwise take weeks to gather manually. With their ability to run on local devices, SLMs offer improved data privacy and reduced infrastructure needs, making them an ideal choice for enterprises. Specifically, more than 75% of organizations are now customizing small open-source models to meet specific requirements, offering a faster, cheaper, and more manageable solution than larger models.
Multimodal AI
Multimodal AI refers to systems that process and integrate multiple types of data—such as text, images, audio, video, and code—within a single model. Unlike traditional AI, which typically handles a single data type, multimodal models can understand and generate more nuanced outputs by reasoning across various inputs. This capability makes them more context-aware, enabling richer, human-like interactions and enhancing creativity and productivity in software development, scientific research, and beyond. As demand for these sophisticated applications grows, McKinsey predicts that multimodal AI will significantly expand over the next 18-24 months, revolutionizing industries by creating more intelligent, adaptable systems.
Is a Large Language Model (LLM) or a Small Language Model (SLM) the right fit for your business? Discover the answer in our article: Breaking Down AI: The Comparative Edge of Language Model Sizes.
More AI advancements in health & life sciences
AI is changing how medical professionals analyze extensive datasets to gain new insights. Its capacity to process and interpret complex information has made AI an essential tool in accelerating disease discovery and advancing drug development.
In 2025, a recent application of AI in life sciences is AlphaFold, an AI system developed by DeepMind. For decades, scientists have struggled with the complex challenge of determining the 3D structures of proteins based on their amino acid sequences—a vital step in understanding how proteins function in the body. This process was traditionally slow, often taking months or even years for a single protein. Thus, Demis Hassabis and John Jumper of Google DeepMind developed AlphaFold to predict the three-dimensional structure of proteins directly from their amino acid sequences with remarkable accuracy.
AlphaFold has made a huge impact on biological research. In the 14th Critical Assessment of Protein Structure Prediction (CASP14), AlphaFold outperformed all other methods. This accuracy allows researchers to model protein structures reliably, which accelerates studies in molecular biology and drug discovery. DeepMind’s open release of AlphaFold’s source code and a database containing predicted structures for over 200 million proteins - including the entire human proteome - has allowed millions of scientists worldwide to access high-quality structural data to expand the coverage of protein structures. With this major breakthrough in AI, the two researchers were awarded the 2024 Nobel Prize in Chemistry.
Another recent breakthrough in 2025 is an AI tool developed by researchers at the University of Pennsylvania that helped identify adalimumab, an FDA-approved monoclonal antibody, as the leading treatment for idiopathic multicentric Castleman’s disease (iMCD). iMCD is associated with cytokine storms, where the immune system releases an excessive amount of inflammatory proteins, damaging tissues and organs and leading to widespread inflammation and potentially life-threatening organ failure. By applying machine learning to analyze 4,000 existing drugs, the research team discovered that adalimumab could effectively counteract the elevated tumor necrosis factor (TNF) signaling in severe iMCD patients. This finding opened the door to a potential life-saving treatment. One patient, who was about to enter hospice care after several failed treatments, was treated with adalimumab and has now been in remission for nearly two years.
These recent advancements have established a strong foundation for the future of AI applications, with the life sciences sector aiming to achieve significant progress in developing treatments for currently hard-to-treat diseases.
See how FPT empowered patient-centric care with AI Solution AIScribe in our latest case study.
Breakthroughs in AI logical reasoning architectures
As business challenges grow increasingly complex, simply retrieving information and generating content is no longer sufficient. AI must evolve to pause, evaluate, and make real-time decisions. Traditional pre-trained models rely on “training-time computing,” which involves predicting outcomes based on vast datasets they have previously encountered. While this approach works for straightforward tasks, it falls short when addressing complicated issues.
This is where AI reasoning becomes crucial. Rather than restating patterns or analyzing past data, AI must engage in “inference-time computing,” where it evaluates various scenarios, considers potential outcomes, and makes decisions from logic. Even though this process demands more effort, it yields significantly more meaningful and impactful results. By empowering AI to “pause and think,” AI can move beyond surface-level responses to tackle the complex problems that drive real business value and innovation.
For instance, OpenAI’s newly released o1 model introduces chain-of-thought (CoT) reasoning, which has significant implications for improving AI accuracy and safety. The key benefit of this approach is that it enables AI to perform real-time double-checking throughout the decision-making process, a feature that helps minimize risks associated with AI hallucinations, biases, and harmful content generation. By embedding this CoT capability into the core operations of o1, OpenAI ensures each step of the AI’s reasoning is reviewed for ethical concerns as it processes information.
AI trends fueled by AI chips
As AI continues to evolve, its advancement depends on a critical foundation: AI chips. As the health and life sciences industry moves forward, AI chips can enable the high-throughput processing needed for genomic data, medical imaging, and real-time diagnostics. Accelerators such as GPUs and TPUs can drive drug discovery and personalized medicine breakthroughs. At the same time, custom AI chips in wearables and medical devices support continuous monitoring and on-device inference.
Similarly, small language models (SLMs) are designed to operate with limited computational resources, making them ideal for low-power edge devices where energy efficiency and quick response times are critical. Custom AI chips enable this by optimizing for fast inference and low latency while consuming minimal power. On the other hand, Multimodal models require AI chips with high memory bandwidth and parallel processing capabilities. These chips can handle the simultaneous data streams and intensive computations needed to deliver real-time, context-aware outputs, making them essential for seamless multimodal experiences.
The rise of AI chips marks a new era in semiconductor innovation, which intensifies competition among tech giants such as Meta, Google, and Intel, as they are developing custom chips to meet growing AI demands. This move is primarily motivated by the need to reduce the high costs of relying on general-purpose processors and external suppliers. Custom AI chips offer businesses enhanced operational efficiency, faster processing speed, and the ability to tailor hardware to specific AI applications. Moreover, owning proprietary chip technology allows companies to better control their infrastructure, improve data privacy, and secure competitive advantages by aligning with their business needs. Meta, for example, has begun testing its in-house AI training chip, part of its broader strategy to reduce infrastructure costs and enhance AI capabilities, with plans to expand usage by 2026. Meta’s latest Training and Inference Accelerator (MTIA) chip significantly improves compute performance and memory bandwidth over its predecessor, supporting complex AI models for recommendation systems and generative AI products.
Vietnam is emerging as a significant hub for AI chip innovation with the launch of FPT AI Factory, equipped with thousands of NVIDIA H100 GPUs - the world’s most advanced AI superchips. Scheduled to begin services in January 2025, this facility offers massive computing power capable of billions of calculations per second, accelerating AI model training and optimization by up to 1,000 times. The factory integrates NVIDIA’s AI Enterprise software stack and supports a comprehensive ecosystem of generative AI technologies, enabling businesses to rapidly create intelligent AI applications and enhance creativity. NVIDIA’s founder, Jensen Huang, emphasizes the importance of local AI development for Vietnam’s industry and people, highlighting the factory’s role in advancing sovereign AI capabilities. Through partnerships with global industry leaders, FPT AI Factory offers early access programs, cloud credits, and expert support to accelerate AI adoption in the region.
The Next Big Things with AI
In the years to come, AI’s impact will continue to grow and offer greater opportunities across all industries. To fully harness the potential of this cutting-edge technology, leading nations, including Vietnam, are focusing on providing domestic talents with AI education, both in universities and the workplace. While AI continues to transform businesses, navigating the evolving landscape can be challenging. Thus, FPT can be a trusted partner in promoting the safe and ethical use of AI to drive digital transformation success for enterprises.
Find more about FPT’s AI service offerings here: https://fpt-aicenter.com/en.
