Brandon Wang is vice president of Synopsys.
The artificial intelligence landscape is undergoing a seismic shift every bit as transformative as the Industrial Revolution and the internet boom. Discriminative AI, including machine learning and deep learning, is reshaping traditional industries while GenAI is starting to push the boundaries of human creativity. When combined with rapid advancements in AI hardware and the promise of emerging technologies like quantum AI, it’s clear we are in just the first steps of an ongoing technological revolution.
AI is relevant across all industries, including the chip design vertical. Now AI-driven electronic design automation (EDA) solutions can deliver over a 10% improvement in performance, power, and area (PPA), up to 10x faster turnaround times and double-digit improvements in verification coverage, among other benefits. In the healthcare industry, AI algorithms can analyze medical images faster and more accurately than human professionals and AI-based drug discovery can significantly reduce the time to market for new therapies.
The Rise Of AI Agents: A New Frontier
One of the most exciting technological developments in the AI landscape is the emergence of AI agents. These function similarly to automated assembly lines, breaking down large tasks into mini-tasks and using AI to execute each one more efficiently. AI agents can be categorized into five levels along a spectrum of autonomy, ranging from simple reactive level L1 agents to fully autonomous L5 agents.
Current agent applications operate mostly at L2 or L3, but the potential for L4 strategic decision-making and adaptive L5 agents could revolutionize sectors from robotics to healthcare. It’s anticipated that L4-level agents will be widely implemented in specialized areas like autonomous robotic surgery by 2035. Applications of L5 AI agents including fully autonomous surgeries and personalized medicine, on the other hand, may not become widespread until 2050 or later.
Among their advantages, these agents promise productivity improvements and may compensate for workforce gaps. However, AI agents also introduce significant challenges. The main hurdle is error compounding, a serious concern in AI systems that perform complex tasks. That’s why higher levels of agent autonomy require increased accuracy to counteract errors that can accumulate in multistage processes. Other challenges include limited context memory where vector databases can cause hallucinations and task planning challenges due to heavy reliance on prompting, which limits scalability.
Transforming The Job Market
The rapid advancement of AI has raised understandable concerns about its impact on the job market. But history suggests that technological disruptions typically increase GDP and create jobs overall, driven by new demands and applications.
A good example is the semiconductor industry, which has been resource-constrained from the start both in terms of capital expenditure and talent. In particular, integrated circuit (IC) designers represent a relatively small talent pool, with an estimated workforce in the tens of thousands globally—significantly less than the 28.7 million software developers and 73 million IT professionals worldwide as of 2024. The demand for IC design work has continually exceeded the available talent, so AI-driven productivity gains can help bridge this gap. For example, if meeting current demand typically requires 100 IC designers, AI could enable 70 designers to achieve the same output, compensating for a workforce shortage or freeing up resources to tackle additional unmet demands. Design automation tools with embedded AI capabilities have proven to increase designers’ productivity.
Insights from McKinsey show the semiconductor industry will need substantial talent investments to meet AI’s growing compute demands, with an emphasis on specialized domains such as AI accelerators and high-performance GPUs. This demand is likely to create tens of thousands of new jobs worldwide over the next decade.
With AI innovation, a long and growing list of new jobs will emerge, such as AI agent developers, ethicists and prompt engineers. In the future, AI personality designers could craft brand-specific AI personalities, AI trainers could enhance models with quality data and fine-tuning and AI system auditors could evaluate for biases and regulatory compliance.
Meanwhile, AI also brings requirements for new skills, and not just technical skills like machine learning and data science. It also requires human-centric skills such as creative problem-solving, critical thinking and emotional intelligence—ones AI won’t be replacing any time soon.
Deploying AI- More Of A Defensive Strategy?
The stakes are high for businesses considering AI adoption, and the competitive risks of delaying implementation are making it more of a defensive strategy. Companies should analyze what kind of impact AI will have on their existing applications, its potential for new applications and technology barriers that can help guide their build-versus-buy strategy.
When deploying AI, businesses should carefully consider timing as well as demand. Would it be better to build internal AI capacity or leverage commercial platforms? Can you satisfy demand with current resources or are opportunities being left on the table?
The Three Waves Of AI Evolution
Major technology disruption is rare and it occurs every two or three decades. The last major phase brought us into the age of the internet, which evolved through three distinct waves. The first was the infrastructure buildout, where foundational support for internet development was established. For example, companies like Cisco and JDSU helped build the networking infrastructure. The second wave brought enterprise-level growth with a focus on developing and managing software platforms and services built on top of the internet technology such as Salesforce and Adobe. The third wave introduced mobility and millions of applications tailored to end consumers’ needs across a range of sectors.
AI is the mega tech disruption now, and it appears to be following a similar path through phases of infrastructure, enterprise and application. So where are we now? Are we at the beginning of the infrastructure wave, in the midst of an explosion of LLMs generated from high-performance computing (HPC) data centers? The demands for semiconductor chips for computing are soaring, whether GPU or custom ASICs. And if so, will the second and third waves of AI—enterprise integration and edge applications—arrive sooner than they did in the internet age? Given the speed of advancements in AI, it’s possible these phases may unfold at a quicker rate, resulting in industries transforming even faster than the internet did.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
