Edge AI: The Next Frontier in Smartphone Innovation
As cloud-based AI faces challenges, edge AI emerges as a promising solution. This shift could revolutionize smartphones, benefiting manufacturers and chip makers while addressing privacy concerns.
Two years have passed since ChatGPT made its public debut, sparking a surge in generative artificial intelligence investments. This trend has boosted valuations for AI startups and tech giants providing cloud computing platforms for AI model training and hosting. However, the current AI boom is showing signs of strain, and the next phase of growth may be found in handheld devices.
Generative AI, which creates new content based on training data, primarily operates in the cloud. For instance, OpenAI utilizes Microsoft's Azure platform for its large language models (LLMs). As these models grow in complexity, so does the infrastructure required to support them.
This has led to a race for building larger and more powerful data centers. OpenAI and Microsoft are reportedly planning a $100 billion data center project set to launch in 2028. Tech giants Alphabet, Microsoft, and Meta Platforms are projected to spend a combined $160 billion on capital expenditures next year, primarily for AI infrastructure.
These enormous investments raise questions about sustainability and return on investment. Companies also face challenges in finding suitable locations for data centers and securing sufficient power supplies. Additionally, the dominance of Big Tech in AI and cloud computing is attracting regulatory scrutiny.
These factors have led to increased interest in "edge AI," which refers to AI algorithms and models that run on devices like smartphones or personal computers. Edge AI offers several advantages:
- Real-time responses without high-speed internet
- Enhanced privacy by keeping personal data on the device
- Potential for rapid adoption due to widespread device usage
Analysts at UBS predict that nearly 50% of smartphones will have generative AI capabilities by 2027, up from just 4% in 2023.
The main challenge for edge AI is technological limitations. Current devices lack the computing power, energy, and memory to run large models like OpenAI's GPT-4. However, there are reasons for optimism:
- Development of smaller, task-specific models
- Improved efficiency of specialized models
- Advancements in semiconductor technology
Leading chipmakers like TSMC, Samsung Electronics, and SK Hynix are pioneering new methods such as advanced packaging in semiconductors. One former TSMC executive predicted that within a decade, this technology could lead to a "multichiplet" containing more than 1 trillion transistors.
Edge AI has the potential to create new winners in the tech industry. While Nvidia's advanced GPUs will continue to dominate, other chip firms like Qualcomm and MediaTek are poised to benefit. The smartphone and PC markets are forecast to surpass $700 billion by 2027, with brands from Apple to Lenovo standing to gain.
"Combined sales in the smartphone and PC markets will surpass $700 billion by 2027, up 14% from this year."
The success of edge AI will ultimately depend on developing compelling applications that users find valuable. If achieved, the next big thing in AI may be found in smaller models and devices, revolutionizing the way we interact with technology in our daily lives.
