It's no secret that the tech industry is in the throes of an AI revolution, with Large Language Models (LLMs) leading the charge. Among the recent converts is VMware, the cloud computing and virtualization technology giant. Despite the VMware's notable contribution to cloud infrastructure and business mobility over the years, their recent strategic focus on AI, specifically LLMs, raises several red flags for me.
To start with, the acquisition of VMware by Broadcom is an unsettling development. Broadcom's track record in software is worrisomely inconsistent, with a focus on short-term profit over long-term innovation. While they pledge significant resources towards R&D, it won't be the VMware R&D we know. The soul of a company often lies in its research and development wing; altering its essence might lead to unpredictable and potentially unwelcome outcomes.
Also, VMware's sudden dive into AI and LLMs appears rather late in the game. Their Silicon Valley neighbors, Microsoft, have already heavily invested to the tune of $10 billion in OpenAI. This investment not only demonstrates a clear understanding of the AI trajectory but also equips Microsoft with a significant head start in both development and implementation. Simply put, VMware is joining a race where competitors are light-years ahead.
VMware has outlined its plans to leverage LLMs across various sectors, namely Customer Operations, Marketing & Sales, Software Engineering, and R&D. While this is undoubtedly an impressive ambition, one cannot help but question their ability to effectively deliver in these areas, given the industry leaders already firmly established in each sector. Giants like Google, IBM, Amazon, and Microsoft have been harnessing AI in these arenas for years, optimizing algorithms, and gaining valuable field experience. VMware are still in the school playground when it comes to AI and LLMs.
VMware's closed-door approach to LLM development also worries me, and it contrasts sharply with any commitment to open-source development (remember Dirk Hohndel?). The open-source model encourages a collaborative approach, harnessing global intelligence to push the boundaries of AI development and implementation . By choosing to buck this trend, VMware could risk sidelining itself from the critical exchange of ideas and methodologies that are shaping the future of AI.
On a relatred note, VMware's private training of LLMs raises industry-wide concerns. Monumental datasets are critical for the training and effectiveness of these models. By keeping such datasets from being openly available, VMware not only limits the scope of their own AI but also hampers overall industry progress. I am not saying it should all be open-source, but where is their contribution to the community here? None, whatsoever.
In a world where AI has become a buzzword, seriously it is getting tedius now, it's become all too easy for companies, and indeed individuals, to tout themselves as 'experts'. The rise of AI platforms such as ChatGPT has resulted in a surplus of self-proclaimed experts who, in reality, have only a superficial understanding of the complexities of AI. A solid background in computer science, data analytics, and machine learning is essential for leveraging the potential of LLMs, and it remains to be seen if VMware can foster or attract this level of expertise, especially with staff leaving like Lemmings.