The Strategic of our CEO’s Guide to Navigating the AI Landscape
This white paper explores the current and future landscape of artificial intelligence (AI), providing CEOs with a comprehensive overview and actionable insights. With the rapid advancements in AI technologies, understanding the implications and opportunities that AI presents for businesses has never been more critical.
First, we delve into the evolution of AI, focusing on the development from monomodal to multimodal foundation models, the distance we have yet to traverse before achieving Artificial General Intelligence (AGI), and the major players in the AI landscape. We also discuss the emergence of novel AI hardware architectures, specifically Intelligence Processing Units (IPUs), and their potential impact on the industry.
Second, we explore the business implications of AI, addressing the concerns of over-dependence on proprietary AI technology and the ongoing global AI investments. We highlight the increasing role of AI in Operational Technology (OT) and the expected cost reductions due to technology advancements.
Finally, we provide actionable advice for CEOs to navigate the AI era effectively. We emphasize the importance of preparing for both the opportunities and challenges that AI brings, encouraging CEOs to focus their investment on value creation, transition from Q&A AI to problem-solving AI, and ensure readiness for the introduction of AI in both IT and OT sectors.
This white paper is a guide for CEOs to better understand and harness the potential of AI in their businesses, thereby positioning their companies to thrive in the era of AI.
As the CEO of a leading corporation, you're already accustomed to making strategic decisions based on technology implications. Today, AI is a critical element of those implications, with a rapid pace of change that promises both opportunities and challenges. This white paper is designed to help you navigate the complex landscape of AI and to make informed, strategic decisions for your organization's future.
The AI landscape is continually evolving. Recently, we've observed significant changes within industry giants like Google and Microsoft. Google's Brain team is undergoing restructuring, with DeepMind taking precedence. Meanwhile, Microsoft is harnessing the capabilities of OpenAI to boost its position in the AI industry.
The global AI community has been closely monitoring Google’s struggle with internal reorganization. As someone who is close to the company, I can attest that the process has indeed resulted in distraction and paralysis. However, it’s important to note that this is the same organization that gave us Transformers, which sit at the root of all the current remarkable advancements in AI. Therefore, it's reasonable to expect that despite their internal turbulence, Google will continue to play a significant role in leading AI technology.
One cannot overlook Microsoft's bold strides in the AI arena. When I spoke to a friend at Microsoft who heads up their AI efforts there, he couldn’t help but express his excitement over how Satya Nadella is aggressively leveraging the partnership with OpenAI. The company is riding the wave of this AI revolution and translating it into near-term revenues, increasing their share in enterprise budgets. However, it remains to be seen if this IT dominance will translate to Operational Technology (OT), where industry leaders are wary of an additional “Microsoft tax”.
However, the AI landscape extends beyond Silicon Valley. Concerns about over-dependence on foreign AI technology have spurred independent AI initiatives in countries like India and Japan. As a result, demand for AI hardware, such as Nvidia AI chips, has surged.
These observations lead us to identify two distinct paths in AI development: global AI-first companies invest heavily in broad, multimodal foundation models, while most enterprises will focus on creating domain-specific AI models. Understanding this dichotomy can help corporations like yours identify opportunities and carve out your unique space in the AI ecosystem.
Foundation models are expanding their capabilities and are becoming increasingly multimodal, which allows for a greater understanding of the world. This understanding spans different modalities like language, images, and video. There's also an important shift in the way we perceive AGI (Artificial General Intelligence). The standard for AGI is no longer necessarily human intelligence. We can foresee AGI developing its unique character, different from human cognition, just as a calculator surpasses human ability in numerical calculations. The potential of AGI is being realized more independently and less tied to comparisons with human intelligence.
While Google and Microsoft are significant players in AI advancements, the recent reorganization of Google's AI efforts and Microsoft's increased focus on OpenAI have stirred the global AI landscape. However, we must consider the geopolitical implications of AI development. Concerns about over-dependence on US AI technology have driven parallel AI investments in countries like Japan, Singapore, Malaysia, and India, similar to their initiatives in the semiconductor and manufacturing sectors.
For example, Japan’s government has determined that AI-independence is a national (economic) security issue. These concerns are asserted more plainly in Japanese-language op-eds than in their English counterparts. Japan has announced multi-billion-dollar investments in projects like Fuguku to ensure national competitiveness and self-sufficiency in this AI era.
Being part of the AI community for many years, I’ve witnessed the evolution of AI hardware from CPU to GPU. Now, the industry is on the brink of transitioning from the current GPU (Graphics Processing Unit) dominated environment to an IPU (Intelligence Processing Units) one. The IPU, which combines compute and memory, is like the biological brain, aligning more closely with the structure of AI computations. The technology for IPUs is still in development, but we expect to see a significant shift towards this new architecture in the next 5-10 years.
As the AI landscape evolved, another significant shift has occurred - the movement of foundation models towards open-source. In the past, these models were proprietary, developed and closely guarded by tech giants. This created a significant barrier for companies wishing to leverage AI, as they had to either partner with these tech giants or invest heavily in developing their own models.
Now, there's a transformative trend underway, with foundation models increasingly becoming open-source. This shift is democratizing access to AI technologies, giving companies the ability to create custom models based on proven architectures. One notable example is Meta's Llama model. Launched as open-source, it has become the progenitor of a rich variety of AI models, forming what can be visualized as a “Llama family tree”.
This chart provides a visual representation of the expansive scope of models that have branched out from the original Llama model. With open-source foundation models like these, it has become far more feasible for companies to build custom AI solutions that are capable and sufficient for their unique needs.
Such open-source models are not mere novelties; they offer the tangible benefit of enabling companies to bypass the substantial costs and efforts involved in developing AI models from scratch. Instead, they can now leverage proven models as a starting point, and focus their resources on customization and application-specific fine-tuning.
This white paper explores the current and future landscape of artificial intelligence (AI), providing CEOs with a comprehensive overview and actionable insights. With the rapid advancements in AI technologies, understanding the implications and opportunities that AI presents for businesses has never been more critical.
First, we delve into the evolution of AI, focusing on the development from monomodal to multimodal foundation models, the distance we have yet to traverse before achieving Artificial General Intelligence (AGI), and the major players in the AI landscape. We also discuss the emergence of novel AI hardware architectures, specifically Intelligence Processing Units (IPUs), and their potential impact on the industry.
Second, we explore the business implications of AI, addressing the concerns of over-dependence on proprietary AI technology and the ongoing global AI investments. We highlight the increasing role of AI in Operational Technology (OT) and the expected cost reductions due to technology advancements.
Finally, we provide actionable advice for CEOs to navigate the AI era effectively. We emphasize the importance of preparing for both the opportunities and challenges that AI brings, encouraging CEOs to focus their investment on value creation, transition from Q&A AI to problem-solving AI, and ensure readiness for the introduction of AI in both IT and OT sectors.
This white paper is a guide for CEOs to better understand and harness the potential of AI in their businesses, thereby positioning their companies to thrive in the era of AI.
As the CEO of a leading corporation, you're already accustomed to making strategic decisions based on technology implications. Today, AI is a critical element of those implications, with a rapid pace of change that promises both opportunities and challenges. This white paper is designed to help you navigate the complex landscape of AI and to make informed, strategic decisions for your organization's future.
The AI landscape is continually evolving. Recently, we've observed significant changes within industry giants like Google and Microsoft. Google's Brain team is undergoing restructuring, with DeepMind taking precedence. Meanwhile, Microsoft is harnessing the capabilities of OpenAI to boost its position in the AI industry.
The global AI community has been closely monitoring Google’s struggle with internal reorganization. As someone who is close to the company, I can attest that the process has indeed resulted in distraction and paralysis. However, it’s important to note that this is the same organization that gave us Transformers, which sit at the root of all the current remarkable advancements in AI. Therefore, it's reasonable to expect that despite their internal turbulence, Google will continue to play a significant role in leading AI technology.
One cannot overlook Microsoft's bold strides in the AI arena. When I spoke to a friend at Microsoft who heads up their AI efforts there, he couldn’t help but express his excitement over how Satya Nadella is aggressively leveraging the partnership with OpenAI. The company is riding the wave of this AI revolution and translating it into near-term revenues, increasing their share in enterprise budgets. However, it remains to be seen if this IT dominance will translate to Operational Technology (OT), where industry leaders are wary of an additional “Microsoft tax”.
However, the AI landscape extends beyond Silicon Valley. Concerns about over-dependence on foreign AI technology have spurred independent AI initiatives in countries like India and Japan. As a result, demand for AI hardware, such as Nvidia AI chips, has surged.
These observations lead us to identify two distinct paths in AI development: global AI-first companies invest heavily in broad, multimodal foundation models, while most enterprises will focus on creating domain-specific AI models. Understanding this dichotomy can help corporations like yours identify opportunities and carve out your unique space in the AI ecosystem.
Foundation models are expanding their capabilities and are becoming increasingly multimodal, which allows for a greater understanding of the world. This understanding spans different modalities like language, images, and video. There's also an important shift in the way we perceive AGI (Artificial General Intelligence). The standard for AGI is no longer necessarily human intelligence. We can foresee AGI developing its unique character, different from human cognition, just as a calculator surpasses human ability in numerical calculations. The potential of AGI is being realized more independently and less tied to comparisons with human intelligence.
While Google and Microsoft are significant players in AI advancements, the recent reorganization of Google's AI efforts and Microsoft's increased focus on OpenAI have stirred the global AI landscape. However, we must consider the geopolitical implications of AI development. Concerns about over-dependence on US AI technology have driven parallel AI investments in countries like Japan, Singapore, Malaysia, and India, similar to their initiatives in the semiconductor and manufacturing sectors.
For example, Japan’s government has determined that AI-independence is a national (economic) security issue. These concerns are asserted more plainly in Japanese-language op-eds than in their English counterparts. Japan has announced multi-billion-dollar investments in projects like Fuguku to ensure national competitiveness and self-sufficiency in this AI era.
Being part of the AI community for many years, I’ve witnessed the evolution of AI hardware from CPU to GPU. Now, the industry is on the brink of transitioning from the current GPU (Graphics Processing Unit) dominated environment to an IPU (Intelligence Processing Units) one. The IPU, which combines compute and memory, is like the biological brain, aligning more closely with the structure of AI computations. The technology for IPUs is still in development, but we expect to see a significant shift towards this new architecture in the next 5-10 years.
As the AI landscape evolved, another significant shift has occurred - the movement of foundation models towards open-source. In the past, these models were proprietary, developed and closely guarded by tech giants. This created a significant barrier for companies wishing to leverage AI, as they had to either partner with these tech giants or invest heavily in developing their own models.
Now, there's a transformative trend underway, with foundation models increasingly becoming open-source. This shift is democratizing access to AI technologies, giving companies the ability to create custom models based on proven architectures. One notable example is Meta's Llama model. Launched as open-source, it has become the progenitor of a rich variety of AI models, forming what can be visualized as a “Llama family tree”.
This chart provides a visual representation of the expansive scope of models that have branched out from the original Llama model. With open-source foundation models like these, it has become far more feasible for companies to build custom AI solutions that are capable and sufficient for their unique needs.
Such open-source models are not mere novelties; they offer the tangible benefit of enabling companies to bypass the substantial costs and efforts involved in developing AI models from scratch. Instead, they can now leverage proven models as a starting point, and focus their resources on customization and application-specific fine-tuning.
