Cloud, Security, and AI, Oh My!
This Week in Enterprise Tech, Episode 11, was about cloud, security, and AI, including a concept called Mixture of Experts, which may be the perfect metaphor for leading the modern IT org.
👋 Hi and welcome to The DX Report — the research hub of The DX Institute all about Digital Transformation, the Digital Experience, and the Digital Enterprise. I’m industry analyst, author, and speaker Charles Araujo, and I’m all about providing insights and analysis for enterprise IT leaders as you make the big bets about your organization’s future!
How do you run and manage an enterprise IT organization today?
There's a pretty good case to be made that it will come down to cloud, security, and AI. All were topics in last week's episode of This Week in Enterprise Tech (sorry, I'm late!).
Of course, it's not quite that simple, is it? While those are three critical top-of-mind issues, they're also underpinned by two foundational concerns that I talk about a lot: business process and data.
And even that's not enough. Because underpinning those critical capabilities are all the functional expertise necessary to manage the network, administer servers (real or virtual), develop apps, manage projects, etc.
The truth is that the organizational and technical complexity of managing an enterprise IT organization far surpasses that of any other function in the enterprise (even if those other functions will never admit it).
So, what's an enterprise leader to do? One of this week's news items offers a clue.
Why You Should Borrow the Mix of Experts (MoE) Concept
In this week's episode, we covered IBM's purchase of Hashicorp, which is all about helping enterprises manage cloud infrastructure. We also covered Thoma Bravo's acquisition of Darktrace, which is all about cybersecurity. And we covered a whole bunch about governing and managing AI. They represent the current triumvirate dominating the enterprise IT conversation: cloud, security, and AI.
But as I just mentioned, there is so much more to manage for an enterprise IT leader, which is why I found one other item so interesting. Snowflake introduced their own LLM (partially in response to Databrick's release several weeks ago), which is notable for its enterprise focus.
However, it also represented a doubling-down within the industry on a newer concept called Mixture of Experts (MoE). The general idea is that rather than continually leveraging a full model with its hundreds of billions of parameters, which is slow and expensive, you can train a bunch of "experts" within the LLM that specialize on a specific domain. Then, after an initial request, the system leverages those "experts" by calling on only the appropriate subset of parameters and get the answer faster, cheaper, and with better context.
While I think this approach is a promising way to evolve and improve LLMs, I also think its a fabulous metaphor for how enterprise leaders can grapple with the onslaught of new and emerging technologies and the resulting complexity that is now essential to IT operations.
We've always done this with functional specialization. But what happens when we take a similar approach with exploring and tackling emerging technologies and looking for ways to apply them within the enterprise? By assigning ownership of the continual research, education, and application of new technologies to various members of your team you can generate excitement and engagement without leaving anyone behind.
Some process and commitment will be necessary, but I think taking this approach could add new life to what may otherwise seem overwhelming. And if you're looking for a place to start, how about by assigning the concept of Mixture of Experts to someone? 😉
Happy listening!
🗓️ This Week in Enterprise Tech, Episode 11
After a long hibernation, the tech market is waking up. IBM and Thoma Bravo make key acquisitions, and Rubrik goes public. Snowflake debuts its new LLM, the DHS reveals the members of its new AI Safety and Security board, and companies continue to overcommit to AI hype. Amalgam Insights’ Hyoun Park and The DX Report’s Charles Araujo sift through This Week in Enterprise Tech.
Segment descriptions and links to all the articles we discuss are in the Show Notes, below.
Watch the full episode here:
Or listen to the episode here: https://www.buzzsprout.com/2319034/14993962
📔 Show Notes
IBM acquired HashiCorp last week for $6.5B. Given IBM’s history of buying companies providing products and services similar to what they already have on hand, whether developed in house or picked up in a previous acquisition, what’s the strategy here? Hyoun and Charles discuss how HashiCorp fits into the IBM portfolio, and what downstream effects will be felt in the multicloud space.
https://newsroom.ibm.com/2024-04-24-I...
In other acquisitions news, Thoma Bravo acquired Darktrace last week for $5 billion, taking it private. The market appetite has been renewed after a quiet time, but the trends are confusing. Charles and Hyoun dig into the specifics, and how it will affect CIOs evaluating mid-size vendors.
https://www.thomabravo.com/press-rele...
Microsoft’s AI advancements are starting to have a noticeable effect on cloud spending, driving more Azure adoption in an environment where AWS has been dominant forever. How can CIOs balance the needs of their AI adopters and AI holdouts? Will the AI golden ticket allow Microsoft to finally catch up to Amazon in cloud adoption?
https://www.reuters.com/technology/mi...
Snowflake recently revealed its new LLM, Arctic. Hyoun and Charles analyze its pragmatic “mixture of experts” approach, and how that will help make AI adoption more manageable and affordable.
https://www.snowflake.com/blog/arctic...
Sifting through AI hype: Cleanlab uses AI to detect AI “hallucinations” - at great expense, where a human expert is still more reliable. Moderna plans to automate all business processes using LLMs. Are there practical lessons to be taken away from these examples? Charles and Hyoun dig in.
https://www.technologyreview.com/2024...
https://www.wsj.com/articles/at-moder...
The Department of Homeland Services announced the formation of its AI Safety and Security Board. Who’s advising the government on AI governance? How will they address AI governance challenges?