Skip to content
November 13 2023

#111 Fine-Tuning in The Sky & AI: A Costly Affair or a Necessity?

Blog Details

There's a buzz in the tech community, allegedly about a possible $3 million cost for enterprises to fine-tune large language models to their specific needs. This hefty figure sparks an intriguing thought. Consider the world of jet engines: when companies like GE supply engines, often through aircraft manufacturers, they aren't offering a one-size-fits-all solution, are they?

Each jet engine, like the CFM56 in commercial aviation, is meticulously customized. Airlines like United or Delta specify their needs – fuel efficiency, thrust, noise control. The manufacturers, including giants like GE, Rolls-Royce, and Pratt and Whitney, expertly adapt these engines, ensuring they excel in diverse flight conditions. This detailed fine-tuning mirrors the precision required in adapting LLMs.

This analogy, connecting AI with aviation, has sparked your curiosity, hasn't it? The maturity of the jet engine market might offer unexpected insights for the evolving world of large language models. What secrets could this established industry unveil about the future of AI? As we embark on this exploratory journey, let’s start with a brief historical dive to uncover these potential parallels

Historical Context and Market Analysis

Forecasting the Future: Winners in the LLM Market

Though the journey of jet engine innovation, spearheaded by France in the early 20th century, saw numerous attempts to commercialize jet engines, the major breakthrough came in the early '70s with Safran Aircraft Engines (then known as Snecma) joining forces with General Electric to create the CFM56 engine, a cornerstone in jet propulsion. Today, this joint venture, CFM International, maintains a strong market presence and holds 39% market share. GE, also has a separate market share of over 10% and that brings the total marketshare of GE family above 50%. 

Drawing parallels to this historical journey in aviation, the development of large language models (LLMs) in the field of AI has its own narrative. The concept of language models has been around for decades in the field of natural language processing (NLP) and computational linguistics. The groundwork for modern LLMs was laid by various researchers and developments in machine learning and NLP over the years including Google's tranformer paper in 2017.

OpenAI's transformative role in the AI landscape is significantly bolstered by its collaboration with Microsoft. This partnership echoes the synergy between Safran and GE in the realm of jet propulsion. Together, OpenAI and Microsoft leverage their distinct capabilities to drive innovation and establish market leadership.

As we've delved into the historical evolution of jet engines and drawn parallels to the development of large language models, let's revisit and expand upon a concept introduced earlier – the 'Generative Seven' or 'G7' of AI. These power players, paralleled in the aviation world, reveal intriguing strategic dynamics in both fields.

The Power Players: G7 of the Jet Aviation – A Deeper Dive

In our previous discussion, we introduced the concept of the 'G7 of Generative AI,' featuring major players like Microsoft, Google/Alphabet, Tesla, Amazon, Apple, Nvidia, and Meta. Building upon this, we now turn our focus to their counterparts in the 'G7 of Jet Aviation.' This comparison between the AI titans and aviation's leading entities offers a unique window into the strategic parallels and potential trajectories shaping the AI landscape.

The aviation industry's jet engine market, much like the rapidly evolving world of AI, is driven by a blend of innovation, competition, and strategic alliances. By exploring these parallels, we gain valuable insights into future strategies and alignments within the AI sector, drawing from the lessons and patterns observed in aviation.

Strategic Parallels

Microsoft and GE: Titans of Innovation

In our aviation analogy, Microsoft's counterpart is unmistakably GE. Both giants dominate their respective fields, blending legacy strength with cutting-edge innovation. GE's prominence in jet engines parallels Microsoft's influence in AI, particularly through its partnership with OpenAI.

Rolls Royce and Google: Catalysts of Technological Revolution

Rolls Royce, a leader in advanced jet engine technology, finds its counterpart in Google, especially in their groundbreaking work in AI. Google's 2017 transformer paper was a pivotal moment in AI, analogous to Rolls Royce's innovations in jet engine technology. Both entities symbolize the spearheading of revolutions in their respective fields.

Pratt and Whitney and Amazon: Diverse Application Specialists

Pratt and Whitney's broad reach in aviation, providing engines and services for diverse aircraft types, mirrors Amazon's wide-ranging influence in AI and technology sectors. Both companies excel in adapting their core expertise to a variety of applications, driving industry-wide progress

NVIDIA and Honeywell: Core Suppliers with Niche Innovations

NVIDIA, in the tech world, parallels Honeywell's role in aviation as indispensable suppliers. Honeywell, while not a dominant force in mainstream jet engine manufacturing, plays a critical role by providing essential components to industry giants such as Boeing, Pratt & Whitney, and Rolls Royce. This aspect of their business underlines their significance as a foundational support to the broader aviation industry.

In addition to this core role, Honeywell also caters to a niche market by producing engines for Light Jets and Very Light Jets, showcasing their ability to balance widespread industry support with specialized product development. NVIDIA mirrors this versatile approach in AI. Best known for GPU technology crucial to a variety of AI applications, they also develop specialized LLMs like ChipNeMo and offer niche AI services through AI Foundations. This multifaceted strategy illustrates a sophisticated blend of providing broad industry support and engaging in targeted advancements, characteristic of both NVIDIA and Honeywell in their respective fields.

Meta and Textron Aviation: Pioneers in Training and Development

Meta, known for its forward-thinking approach, finds its aviation equivalent in Textron Aviation. Just as Textron supports pilot training through Cessna, Meta is carving out its niche in developing foundational AI technologies.

Tesla and Honda: Pioneers in Niche Innovation

In the realm of AI, Tesla's role is characterized by its focus on pioneering and specialized applications, a strategy mirrored in aviation by Honda. Honda's entry into the light jet market with its innovative engine designs represents a similar approach to Tesla's in AI - venturing into specialized, groundbreaking domains. Both entities exemplify the importance of innovation in niche sectors, highlighting the significant contributions that specialized focus can bring to both the AI and aviation industries.

As we've explored the strategic parallels between AI companies and jet aviation, our journey now takes us to a broader industry perspective. We'll delve into how the aircraft manufacturing giants mirror cloud computing leaders, examining the strategies and market positions of these industry behemoths. This comparison between the aviation and cloud computing sectors will offer us further insights into the evolving dynamics of these specialized industries.

Broader Industry Perspectives

The Giants of the Skies: Aviation Meets Cloud Computing

Just as the cloud computing world is primarily led by AWS and Azure, the aircraft manufacturing industry sees a similar scenario with Boeing and Airbus at the helm. These two juggernauts are to aviation what AWS and Azure are to cloud computing – dominant and pervasive. However, in this landscape, there's also Embraer, akin to Google Cloud Platform (GCP) in the cloud sector. Embraer, while not as large as Boeing and Airbus, plays a critical role, especially in the regional aircraft market, much like GCP's innovative and significant presence in cloud computing. This parallel not only highlights the diversity within these industries but also underscores the importance of both major and emerging players in shaping the market.

Looking Ahead: AI and Aviation's Shared Horizon

As our exploration of AI and aviation concludes, we're left with profound insights into their intertwined futures. The journey of AI, paralleling aviation's evolution, highlights a dynamic, collaborative landscape far from a winner-takes-all scenario. This parallel not only illuminates the present state of AI but also its expansive potential.

The emergence of smaller-form-factor LLMs and very light jet markets in both fields underscores an industry characterized by adaptability and innovation. These advancements promise to democratize AI, broadening its reach and impact.

As we look forward, it's clear that AI, like aviation, is poised to profoundly reshape our world. This journey, rich in innovation and collaboration, invites us to envision a future where technology and human experience converge, soaring to new heights and redefining our societal landscape.