Intel: The Nokia in the AI Era
Intel's recent announcement of integrating AI, specifically language models, into every platform has stirred both excitement and skepticism. CEO Pat Gelsinger's commitment to building AI into every platform, as detailed in this article from The Verge, highlights the company's ambitious direction towards AI enablement. While the integration of neural computing into chip design offers potential advantages, the push for local language models raises questions. This article critically examines Intel's current strategy, drawing parallels with its recent history of prioritizing legacy business protection, and calls for a change in mindset to embrace the new era of computational hardware.
Intel's strategy to incorporate neural computing into chip design is not without merit. Applications such as real-time data processing and enhanced security can benefit from this approach. However, the push for local implementation of language models brings associated costs and limitations. Intel is not alone in this pursuit; others in the industry are exploring edge-based LLMs, as discussed in this Forbes article and its follow-up piece. These articles project significant growth in data center infrastructure and operating costs, leading to a push for on-device computing. Yet, the question remains: do the benefits outweigh the associated costs and limitations?
The push for on-device AI computing, as highlighted in the Forbes articles mentioned earlier, emphasizes user steering, optimization in performance, efficiency, cost-effectiveness, offloading work for GenAI, and power consumption limits. However, a critical examination reveals that on-device computation may not be suitable for performance-critical tasks like GenAI.
While GenAIs are mostly performance-critical, on-device computation is not designed for such demanding tasks. If total power consumption, performance, and efficiency are to be concerned as GenAI develops further, the trend should accelerate towards migrating more computation into the cloud, not the other way around. User steering at the edge can provide flexibility, but cloud computation should still handle AI training and optimization to meet market demands for efficiency, performance, and power.
The push for Edge AI or on-device AI computing for LLM may prove to be only a concept, lacking practical application in real-world scenarios. Intel's strategy, aligning with this trend, may be misguided, failing to recognize the inherent limitations and contradictions of on-device AI for performance-critical tasks.
The competitive landscape of computational hardware is rapidly evolving, with companies like AMD and Nvidia leading with innovative approaches. AMD's recent reveal of a new AI chip to challenge Nvidia's dominance, as reported in this CNBC article, underscores the fierce competition and relentless innovation in the AI hardware space.
Intel's absence from this front-line competition is astonishing. While AMD and Nvidia are pushing the boundaries of AI hardware, Intel seems to be protecting its legacy products by imagining a future of edge computing in AI. This strategy, while grounded in Intel's traditional strengths, may not be sufficient to compete in this new era. The industry demands agility, creativity, and a willingness to take calculated risks.
Intel's focus on edge computing for AI, particularly for performance-critical tasks like GenAI, appears misaligned with the market's direction. As competitors forge ahead with cutting-edge AI hardware, Intel's strategy seems to lag, clinging to concepts that may prove to be impractical in real-world scenarios.
The question arises: Is Intel's strategy a visionary move or a misguided attempt to protect legacy business at the expense of true innovation? The answer may determine Intel's position in the rapidly changing landscape of computational hardware.
Nokia's downfall is a well-documented case study in the failure to adapt to a rapidly changing market. Once a leader in mobile technology, Nokia's decline offers striking parallels with Intel's current situation. Here's how the two stories align:
Resistance to Change: Nokia's reluctance to embrace the smartphone revolution, sticking instead to its traditional feature phones, mirrors Intel's clinging to legacy products and concepts like edge computing for AI. Both companies have shown a tendency to protect existing business models rather than innovate and adapt.
Ignoring Market Trends: Nokia failed to recognize the rise of touch-screen interfaces and app ecosystems, leading to a loss of market share to competitors like Apple and Android. Similarly, Intel seems to be missing the front-line competition in AI hardware, focusing on concepts that may not align with industry demands.
Lack of Agility: Nokia's bureaucratic structure hindered its ability to respond quickly to market changes. Intel's focus on edge computing for performance-critical tasks like GenAI, despite apparent contradictions and limitations, may reflect a similar lack of agility and responsiveness to the competitive landscape.
Misplaced Investments: Nokia invested heavily in its Symbian OS, even when it was clear that it couldn't compete with iOS and Android. Intel's investment in local language models and on-device AI computing may prove to be a similar misstep if these concepts fail to materialize in practical applications.
Failure to Leverage Strengths: Nokia had significant strengths in hardware design and global reach but failed to leverage them in the new era of smartphones. Intel's strengths in chip design and manufacturing could be similarly underutilized if the company continues to pursue strategies that don't align with the direction of the AI hardware industry.
Complacency: Perhaps the most striking similarity is a sense of complacency. Nokia's dominance in the mobile phone market may have led to overconfidence and a failure to see the threats on the horizon. Intel's legacy success and dominance in certain markets may be leading to a similar complacency, ignoring the innovations of competitors like AMD and Nvidia.
The lesson from Nokia's downfall is clear and urgent: adapt or become irrelevant. Intel's current strategy, with its focus on edge computing for AI and protection of legacy business, risks repeating Nokia's mistakes. The analogy between the past Nokia and today's Intel serves as a stark warning and a call to action.
Nvidia's CEO, Jensen Huang, recently referred to chatGPT as the "iPhone moment" for AI, as reported in this article from ExtremeTech. This statement captures the transformative potential of AI and the excitement surrounding its continued evolution. The question now is, who will become the new Nokia?
Although the title of Intel's announcement bears similarity to Nvidia's announcement, the underlying message is clear: Intel is not in the competition. While Nvidia and others are forging ahead, embracing innovation and shaping the future of AI, Intel's focus on legacy protection and edge computing may leave it trailing behind.
The analogy between Nokia's past and Intel's present is more than a cautionary tale; it's a stark reflection of the choices and challenges facing Intel today. As the AI era unfolds, Intel's strategy and decisions will determine whether it leads, follows, or becomes the Nokia of the AI era.
(Update Sept. 1st, 2023: According to this OpenAI blog, “….we’ve seen teams adopt it in over 80% of Fortune 500 companies”. Intel, most likely, will prove that their local LLM approach is at the wrong side of the ecosystem, and will lead the company to become the Nokia in AI era)
(Update Jan 4th, 2024: Intel to spin out AI software firm with outside investment. It seems that Intel is trying to be in the game of AI. We will see later how this plays out. )
(Update March20, 2024: Biden-Harris Administration Announces Preliminary Terms with Intel to Support Investment in U.S. Semiconductor Technology Leadership and Create Tens of Thousands of Jobs )
(Update 4/23/2024: Intel completes assembly of first commercial High-NA EUV chipmaking tool — addresses cost concerns, preps for 14A process development in 2025. I sincerely wish Intel succeed in their endeavor of #buildInAmerica )
(Update Aug 1st, 2024: Chipmaker Intel to cut 15,000 jobs as tries to revive its business and compete with rivals . It’s distressing to witness this trend unfolding. One fundamental step they could have taken is to reduce these jobs before shifting their focus to the foundry business. Clearly, the federal funding they are receiving from the CHIPS Act will be squandered by the incompetent management.)
(Update March 12, 2025: After Intel fired their CEO Gelsinger in Dec 2024, they hired Lip-Bu Tan as their new CEO today. We will see how Tan will “reinvent” Intel)
(This article was written, NOT to criticize Intel, but for my great grandchildren’s entertainment purpose, so please read it at your own risk. With the help of ChatGPT)