ai companies value hardware solutions inference also

Artificial intelligence: The time to act is nowArtificial intelligence will soon change how we conduct our daily lives. Arecompanies prepared to capture value from the oncoming wave of innovation?Pity the radiology department at your local hospital. Yes, they have a fineMRI machine and powerful software to generate the images. But that’s where themachines bog down. The radiologist has to find and read the patient’s file,examine the images, and make a determination. What if artificial intelligence(AI) could jump-start that process by enabling real-time and more accuratediagnoses or guidance, beyond what human eyes can see?Thanks to technological advances over the past few years, manufacturers areclose to offering such leading-edge MRI solutions. In fact, they’re exploringnew AI applications that span virtually every major industry, from industrialsto the public sector. With better algorithms and increased stores of data, theerror rate for computer calculations is now often similar to or better thanthose of human beings for image recognition and several other cognitivefunctions. Hardware performance has also improved drastically, allowingmachines to process this unprecedented amount of data. That has been a majordriver of the improvement in the accuracy of AI models.### The evolution of AIArtificial intelligence (AI) was born in the 1950s, when the English polymathAlan Turing created a test to determine a machine’s ability to mimic humancognitive functions, including perception, reasoning, learning, and problemsolving. AI grew with the rise of machine learning (ML)—wherein systems absorband “learn” from data. They then use this knowledge base to make betterpredictions and decisions over time. In 2010, the advent of deep neuralnetworks ushered in the deep learning (DL) era.All ML and DL solutions require two steps: training and inference. Take thesoftware in autonomous cars. To help systems detect obstacles in the road,developers present images to the neural net—for instance, those of dogs orpedestrians—and perform recognition tests. Network parameters are then refineduntil the neural net displays high accuracy in visual detection. After thenetwork has viewed millions of images and is fully trained, it enablesrecognition of dogs and pedestrians during the inference phase.Training now accounts for about 95 percent of AI-related workloads in thepublic cloud because most AI applications are still relatively immature andrequire huge amounts of data to refine them. As AI models mature, inferencewill gain more share in the cloud. In fact, DL inference could account for 30to 40 percent of public-cloud workloads over the next three to five years,with training dropping to 60 to 70 percent. Inference will also gain sharewith the rise of edge computing (which takes place within devices), asinnovation enables low-power, high-performance inference chips.Within AI, deep learning (DL) represents the area of greatest untappedpotential. (For more information on AI categories, see sidebar, “The evolutionof AI”). This technology relies on complex neural networks that processinformation using various architectures, comprised of layers and nodes, thatapproximate the functions of neurons in a brain. Each set of nodes in thenetwork performs a different pattern analysis, allowing DL to deliver far moresophisticated insights than earlier AI tools. With this increasedsophistication comes greater needs for leading-edge hardware and software.Well aware of AI’s massive potential, leading high-tech companies have takenearly steps to win in this market. But the industry is still nascent and aclear recipe for success hasn’t emerged. So how can companies capture valueand see a return on their huge AI investments?Our research, as well as interactions with end customers of AI, suggests thatsix tenets will ring true once the dust settles. First off, value capture willinitially be limited in the consumer space, and companies will achieve mostvalue by focusing on enterprise “microverticals”—specific use cases withinselect industries. Our analysis of the technology stack also suggests thatopportunities will vary by layer and that the most successful companies willpursue end-to-end solutions, often through partnerships or acquisitions. Forcertain hardware players, AI might represent a reversal of fortune, afteryears of waning interest from investors who gravitated toward software,combined with heavy commoditization that depressed margins. We believe thatthe advent of AI opens significant opportunities, with solutions in both thecloud and the edge generating strong end-customer demand. But our mostimportant takeaway is that companies need to act quickly. Those that make bigbets now and overhaul their traditional strategies will emerge as the winners.Edge and cloud solutionsOur core beliefs about the future of AI1. Value capture will initially be limited in the consumer sector2. Enterprise winners will focus on microverticals in promising industries3. Companies must have end-to-end solutions to win in AI4. In the AI technology stack, most value will come from solutions orhardware5. Specific hardware architectures will be critical differentiators for bothcloud and edge computing6. The market is taking off already—companies need to act now and reevaluatetheir existing strategies(…)Nvidia’s success shows that tech companies won’t win in AI by maintaining thestatus quo. They need to revise their strategy now and make the big betsneeded to develop solid AI offerings. With so much at stake, companies cannotafford to have a nebulous or tentative plan for capturing value. So what aretheir main considerations as they forge ahead? Our investigation suggests thefollowing emerging ideas on the classic questions of business strategy: * Where to compete. When deciding where to compete companies have to look at both industries and microverticals. They should select the use cases that suit their capabilities, give them a competitive advantage, and address an industry’s most pressing needs, such as fraud detection for credit-card transactions. * How to compete. Companies should be searching now for partners or acquisitions to build ecosystems around their products. Hardware providers should go up the stack, while software players should move down to build turnkey solutions. It’s also time to take a new look at monetization models. Customers expect AI providers to assume some of the risk during a purchase, and that could result in some creative pricing options. For instance, a company might charge the usual price for an MRI machine that also has AI capabilities and only require additional payment for any images processed using DL. * When to compete. High-tech companies are rewarded for sophisticated, leading-edge solutions, but a focus on perfection may be detrimental in AI. Early entrants can improve and rapidly gain scale to become the standard. Companies should focus on strong solutions that allow them to establish a presence now, rather than striving for perfection. With an early success under their belt, they can then expand to more speculative opportunities.(…)If companies wait two to three years to establish an AI strategy and placetheir bets, we believe they are not likely to regain momentum in this rapidlyevolving market. Most businesses know the value at stake and are willing toforge ahead, but they lack a strong strategy. The six core beliefs that we’veoutlined here can point them in the right direction and get them off to asolid start. The key question is which players will take this direction beforethe window of opportunity closes.###### About the author(s)More: McKinsey