If there’s a leading technology of the current era, artificial intelligence (AI) is clearly a top contender. The hype is constant and flows from all quarters. AI’s role in consumer products and enterprises alike is growing, rare for any technology.
AI as a platform spans hardware, software, and on-demand services. All three categories have very different players, although there is some overlap between hardware and software players.
The number of U.S. AI companies has doubled since 2017. According to Tracxn Technologies, which tracks startup businesses, as of the third quarter of 2022, there are 13,398 artificial intelligence startups in the United States.
IDC predicts the worldwide AI market, including software, hardware, and services, will grow from $327.5 billion in 2021 to $554.3 billion in 2024 with a five-year compound annual growth rate (CAGR) of 17.5%.
Also see: What is Artificial Intelligence
What is AI?
Because it is so widely used, AI has become tricky to define. Ask ten people to define AI and you will likely get ten different variations. IT consultancy Gartner defines it as “applying advanced analysis and logic-based techniques, including machine learning (ML), to interpret events, support and automate decisions and to take action.”
This definition coincides with the current state of AI technologies, which is to say that AI performs data analysis and conducts actions based on the findings of that analysis. Analytics has been around for quite some time but it tends to use smaller data sets and offers smaller, less elaborate results. AI handles larger datasets and offers more probabilistic outcomes.
Clearly, there is a wide range of ways in which artificial intelligence can be used. Not to mention the many subdivisions of AI, such as machine learning, deep learning, and natural language processing (NLP).
Therefore, it is incumbent on customers of AI vendors to ask them their definition of AI, and how their offerings will meet your expectations. You have to make sure that their vision of AI aligns with your business needs.
Also see: Top AI Software
According to the market research firm Tractica, the global AI-driven hardware market is in the process of growing from a mere $19.63 billion back in 2018 to an expected $234.6 billion by 2025. The AI-driven hardware market includes categories such as CPU, GPU network products and storage devices.
When it comes to AI hardware, the real players are chipmakers, because AI processing is vastly different from typical application processing using CPUs. For the most part, that involves GPU makers, but in recent years there have been startups using new chip designs specifically geared toward AI processing in the hopes of being more efficient and faster than GPUs.
The leading vendor in AI hardware is GPU maker Nvidia. It has repurposed chips normally used to accelerate video games as AI processors, working much faster than an x86 CPU. AMD was not much of a player in this field for a long time because it was struggling to survive, but it has made a remarkable comeback in recent years and is now making serious inroads in the AI and high-performance computing (HPC) market.
Intel is also finally finding its footing in the AI space. It has an inference processor, called Goya, and a processor specifically for self driving cars called Mobileye, as well as its Altera FPGA line for training processing. But it never could quite get the GPU product right until now. Its Xe architecture will be sold under the Arc brand name for consumer GPUs while the AI/HPC product will be known as Ponte Vecchio.
All of the major server vendors – the top brand names like HPE, Dell, and Lenovo as well as vendors such as Supermicro, Wiwynn, and Inspur – all have AI-oriented hardware using chips from Intel, AMD, and Nvidia.
Also see: AI vs. ML: Artificial Intelligence and Machine Learning
Getting an accurate measure of the overall AI software market is challenging, because many general purpose applications have AI in them – this leads to the question of whether they should be considered AI software or merely software with AI capabilities.
If we go with the latter, the overall market is massive because it breaks down into so many different categories, each with multiple competitors.
And it’s not always who you might think is a leader. One of the biggest AI software vendors is Nvidia, which we’ve already mentioned in the hardware category. The company often boasts that it has more software engineers than hardware engineers, making AI software to run on their GPUs.
But there are many other vendors. Gartner estimates worldwide AI software revenue was $62.5 billion in 2022, an increase of 21.3% from 2021.
Also see: The Future of Artificial Intelligence
AI is also being made available as a service, just like software, infrastructure, platform, and other on-demand services through cloud service providers. AI-as-a-service has an appeal to many midsized and smaller enterprises because it means that they don’t have to make the massive investment in AI hardware.
AI hardware is extremely powerful. It’s also extremely expensive. The only real need for horse power is in the training segment. The inference portion of AI, which is where it will mostly be used, does not require high performance computing. A company may perform algorithm training just a few times a year, but then run inferencing against those algorithms as part of business.
That means a company’s expensive AI training hardware, which can easily run into the six and seven figures, will sit idle for long periods because it’s not needed. So why buy when you can rent for the short period you need it? Using AI-as-a-Service, a company on a budget can do the expensive training portion through a cloud provider for much less than the cost of investing in the hardware.
AI-as-a-Service is provided by the top cloud hyperscalers: AWS, Microsoft, Google, and in particular IBM. IBM has lagged behind the other major cloud vendors in overall cloud market share, but it has made a significant AI effort with IBM Watson Cloud. First, it allows companies to make AI a part of their existing applications to make more accurate predictions, automate the decision making processes, and get optimized solutions.
Watson has a number of pre-built applications, such as Watson Assistant, Watson Speech to Text, and Watson Natural Language Understanding. IBM Watson Cloud also provides AI solutions for specific markets such as AI for Customer Service, AI for Financial Services, and AI for Cybersecurity.
Also see: How AI is Altering Software Development with AI-Augmentation
Toward an AI Strategy
For a business to truly gain the benefits of AI, it should be deployed enterprise-wide, because the benefits of AI can be most fully realized across virtually every department in the company. Gartner says a proper AI strategy identifies use cases, quantifies benefits and risks, aligns business and technology teams and changes organizational competencies to support AI adoption.
The first step is to focus on what your organization is trying to accomplish and the business problems you’re working to solve. AI does not have to be for new applications only. It can be made part of your existing suite of applications, as IBM is trying to do with Watson.
However, it should be done slowly with great deliberation, for two reasons: first, there’s a learning curve inherent in every new technology. No matter how talented the IT staff, they are still going to need time to grasp all of the fundamentals of AI programming and integration in with their applications.
The second reason is that Gartner has noted that organizations experiment with AI but often struggle to make the technology a part of their standard operations. That’s because AI is still in its early stages and the maturation process cannot be rushed. Gartner predicts that it will take until 2025 for half of organizations worldwide to reach what Gartner’s AI maturity model describes as the “stabilization stage” of AI maturity or beyond.
The Future of AI
The future of looks to be more, faster, and larger investments. Clearly there will be many more use cases for AI, many more applications. The hardware will get faster, leading to more powerful AI systems. And the data sets will get bigger, meaning more complex AI applications in the future.
Beyond the usual speeds and feeds will be the next big step in AI, known as artificial general intelligence, or AGI. Whereas AI does what is programmed to do, AGI has initiative. It asks questions that were not part of its programming and acts upon the.
That may frighten some people. From “2001: A Space Odyssey” to “The Matrix,” fear of AI coming to life and turning on humanity has been around for decades. Hopefully, the benefits will be obvious and overcome any fears induced by popular culture.