This new blog series explores artificial intelligence and automation in technology and the important role it plays in the Broadcom portfolio.
This Easter, I commissioned text-to-art AI tool Midjourney to create a futuristic egg basket that showcases the concept of digital connectivity. What I saw blew my mind. The artwork I received was not only visually stunning, but also showed how AI can bring new ideas to life. As a creator, I used Midjourney, chatGPT, and other tools to experience first-hand the transformative nature of generative AI.
AI is now rapidly becoming pervasive. The world is waking up to the power of generative AI, and an entire ecosystem of applications and tools is rapidly going live. Its importance is increasing in various industries such as healthcare, finance and transportation.
All of these have had a tremendous impact on the digital value chain and semiconductor hardware market and cannot be ignored. Apps and tools should collect, process, and send data back to consumers with minimal delay. Given the rapid emergence of ecosystems, this also needs to be done at scale.
Hardware innovation will be essential to sustain this revolution. As a content creator using Midjourney for art and chatGPT for editorial assistance (as in this blog) when I think about the growth of AI and its impact on hardware systems and silicon roadmaps. Our journey helps clarify the big picture.
So what do you need on the hardware side?
Growing demand for high-performance computing for cloud data centers
AI workloads require specialized processors capable of processing complex algorithms and large amounts of data. This has led to the development of new processors optimized for AI workloads, including graphics processing units (GPUs), field programmable gate arrays (FPGAs), and custom AI silicon.
memory and storage
The sheer volume of data generated by AI workloads requires high-capacity storage solutions that can handle both structured and unstructured data. Solid State Drives (SSD) and Non-Volatile Memory Express (NVMe) enable faster data access and processing.
Wired connections as binding threads
Computing speed, loosely measured as a function of Moore’s law, cannot keep up with the exploding data needs from AI. This means that interconnections between computing units working in parallel become important. Wired connectivity plays a vital role between computing devices, GPUs, storage and memory. The rise of AI means that networks are computers and connectivity is their lifeline.
wireless at the edge
It goes without saying that our digital experience today is wireless. To achieve this goal, it is critical that wireless broadband networks can handle high-speed, low-latency data at scale. Cellular technology and Wi-Fi work complementary to meet this demand.
The introduction of 5G will continue to increase the capacity of cellular networks and provide the highest speeds. Wi-Fi innovation culminates with Wi-Fi 7, which, combined with the newly available 6 GHz band, directly addresses edge latency issues while ensuring top speeds.
cyber security
With the increasing use of AI, cybersecurity is becoming more and more important to protect against cyberthreats and attacks. This requires specialized hardware and software such as intrusion detection systems and encryption technology.
At Broadcom, we’ve always aimed to connect everything. That is our mission. Connecting the world with Broadcom. We are very proud of the fact that 99% of the world’s data passes through at least one Broadcom chip. For us, our AI hardware needs lie in the continuity of what we do every day. Our wired connectivity powers hyperscaler data centers and operator networks. At the edge, our wireless broadband solutions power and deliver data to your home. Our software solutions provide the security layers you need. We welcome AI to disrupt the status quo and are proud to provide the critical hardware that makes this possible.
Finally, let me give you a quick preview of some of the innovations we’re focusing on.
Advanced manufacturing process: These are essential for manufacturing AI hardware. Using 7-nanometer and 5-nanometer manufacturing processes creates smaller and more powerful chips. Our chips carry higher data loads and deliver the low latency needed for AI.
Custom design: We innovate data center storage and connectivity solutions optimized for specific AI workloads.
Power efficiency: Complex workloads require large amounts of power, which among other things can lead to increased energy costs. This is an area where both our wired and wireless chips are focused. For example, Wi-Fi chips used in phones are resolutely committed to radio optimization and architectural changes with each generation, with the goal of significantly reducing power consumption.
These are just a few examples of the innovations we focus on. In a series of follow-up blogs, we hope to delve even further into what Broadcom can offer as AI becomes more prevalent. Get an update on how AI will shape the software and semiconductor hardware markets and how our innovations will meet the demands of AI applications.
Learn more about Broadcom innovations here.
