Stewart Squared
Episode #44: Consultants Preach, AI Learns, and CEOs Fall Behind
- AI & Agents
- consultants in tech
- business school leadership
- technical vs non-technical CEOs
- McKinsey history
- Webvan failure
- Amazon Web Services
- CIA cloud contract
- Oracle vs Amazon lawsuits
Listen
Download MP3About this episode
Welcome to Stewart Squared podcast with the two Stewart Alsops. In this episode, they explore why consultants often fail in the tech world, how leadership skills are (or aren’t) taught in business schools, and the historical tension between technical and non-technical CEOs. They trace the evolution of Silicon Valley’s culture, from the idealistic hackers of the PC revolution to Amazon’s strategic rise with AWS and its CIA contract, and discuss whether institutional knowledge should be centralized or decentralized inside corporations. The conversation ranges from the origins of corporations and supply chain mastery at Apple, to predictions with LLMs, IoT security challenges, and even why Google struggles to innovate beyond its search monopoly. Show notes include a recommendation to read Apple in China for deeper insight into Apple’s role in training millions of Chinese factory workers.
Check out this GPT we trained on the conversation
Timestamps
00:00 – Opening with Stewart Alsop III teasing topics like why consultants fail in tech and the theory that post-founder CEOs rarely succeed, leading into the history of McKinsey and the Big Five consulting firms.
05:00 – Critique of MBA programs for focusing on analysis over leadership, discussion of Stanford GSB and Harvard HBS networks, and whether leadership can be taught.
10:00 – Exploration of technical vs non-technical CEOs in Silicon Valley, examples like Steve Jobs, Larry Ellison, and the early PC industry’s bias against consultants.
15:00 – Deep dive into Amazon Web Services, Andy Jassy’s startup-first strategy, and AWS’s CIA cloud contract, plus Oracle’s legal battles over DoD’s JEDI contract.
20:00 – Debate on AI prediction limits, the MIT SEAL framework for updating LLM weights, and real-time adaptability in AI models.
25:00 – Examination of corporations as knowledge bodies, historical roots in Dutch East India Company, and the tension between centralized vs decentralized knowledge.
30:00 – Focus on institutional memory, Apple’s supply chain with Tim Cook, United Airlines’ IT transformation, and IoT security risks.
35:00 – Insights on device authentication, Device Authority’s IoT security approach, and vulnerabilities like Stuxnet.
Key Insights
- Consultants often fail in tech leadership because they lack deep domain expertise and tend to focus on analytical frameworks over practical execution. The Alsops argue that consultants are great at creating presentations and identifying what companies should have done but struggle to navigate the messy realities of running large, complex organizations—highlighted by the Webvan example where a consultant-turned-CEO helped drive the company into bankruptcy.
- Business schools train analysts, not leaders, equipping graduates with skills in spreadsheets, case studies, and presentations rather than fostering the hands-on leadership required in startups and tech firms. While MBAs can be valuable for networking and strategy roles, they often fall short in preparing executives to scale companies or inspire teams in rapidly changing environments.
- Technical and non-technical CEOs shape companies differently, with early Silicon Valley favoring technical founders like Gates and Wozniak. However, leaders like Steve Jobs and Larry Ellison thrived without deep technical skills by surrounding themselves with strong technical co-founders, showing that vision and communication can sometimes outweigh engineering chops in the CEO role.
- Amazon’s AWS strategy illustrates effective knowledge transfer and scaling, starting with a focus on startups and evolving to win contracts like the CIA’s cloud infrastructure. Andy Jassy’s ability to scale AWS from an internal tool to a dominant cloud service underscores how decentralized initiatives can later become centralized strengths when aligned with leadership vision.
- The SEAL framework represents a breakthrough for LLMs, enabling models to update their weights post-deployment for real-time learning. This adaptation could blur the line between static and dynamic AI systems and marks an early step toward meta-learning, raising both exciting possibilities and existential concerns about machine autonomy.
- Institutional knowledge must balance centralization and decentralization. Centralized databases simplify operations, as seen in United Airlines’ customer system, but decentralized human knowledge prevents organizations from collapsing when key people leave. Apple’s reliance on Tim Cook as its operational brain is cited as both a strength and a cautionary tale about knowledge bottlenecks.
- IoT security remains a critical and under-addressed challenge, with billions of devices running outdated software and exposing organizations to risk. Companies like Device Authority are working on real-time device identification and updates, but widespread implementation lags, creating vulnerabilities that even nation-state hackers have exploited, as in the Stuxnet incident.
Episode transcript
Everyone is wrong about AI, but if you know the history of technology, you can be wrong and still be right.
I know the history of technology since I started with the personal computer in 1980. So listen in while we explore what is right about AI.
Welcome to the Stuart Squared podcast. We've got a great set of talking points that we will completely ignore as we follow whatever train of thought is most interesting at the time. But the two first ones are that you've developed a theory that the CEO appointed after the founder in a big company never works out. And the other interesting talking point is about consultants and why consultants never work out. But let's start with the history of consultants. When did like McKinsey and all the Big Five become a thing? And why is there so much tension between them and the tech world?
Let's constrain that just to the tech industry because you know the big consultants and I'm not sure there are five of them left, but the big consultants have a range of consultant business that they do in different industries. But in the tech business there's this distinction in tech, which is in order to be a successful CEO, you have to understand the technology and consultants are trained to be independent of any particular industry, not have any particular domain knowledge. It breaks down inside of a consultancy that certain people develop expertise in a particular domain, but they're consultants and they're not running a company. So their job is to provide advice. So the first level of this is my own bias about consultants, which I'll make clear, which is consultants tell you to do what you should have done in the first place. So consultant doesn't solve the problem of why a company doesn't do it in the first place. Usually that's a result of it's gotten really big, it has multiple kinds of businesses, it's and its sort of mission has become diffuse. So you have to hire consultants to tell you what to do with this business or that business. So if you look at MBA programs and the two kind of top of mind are Stanford, gsb, Graduate School of Business or Harvard. What the hell is called business class? Hbs. So Harvard, the joke is Harvard put the BS in hbs. There's no equivalent joke for Stanford, but these are the two that are kind of considered top of line. And there's sort of three reasons people go to business school who know why they're going to business school. Plenty of people don't. And one is in your graduating class in business school. These are the best of the best, right? So you start out your career if you do business school right after your undergrad, knowing the people who are going to be running stuff going forward. And so it's a great network to have. This is particularly true at Stanford, particularly in technology companies. But the second thing is that business schools train you to do something particular and it's really all about analytics analysis. So Harvard's famous for doing case studies and they'll study something, document it, write these case studies kind of independently of the classes. And the professors use them to teach classes, depending on what they're teaching. And there are some of these cases that are truly famous. One I remember, I can't give you the details is a company was making growing potatoes and wanted to sell French fries to McDonald's. And so they studied McDonald's in detail and discovered that McDonald's had French fries that are a particular width. So they optimized for once they'd grown the potato for making exactly what McDonald wants and then set up a distribution system for them. So, you know, get to know your customer kind of case study. There's others. And so the third point is that a lot of what you learn as an analyst in business school, it's not necessarily practical day to day issues of running a company as much as it is analyzing companies to figure out what the problem is or how to do more of the successful thing. And so a lot of the skills you learn are how to create slide presentations, how to use spreadsheets to understand numbers, all those sort of gory business stuff, which is not what teaches you to be a leader. And MBAs. And maybe they've gotten more into this now, but certainly in my lifetime sort of missed the point about teaching leadership. And can you teach leadership and stuff like that?
Your first full transcript is free. After that, an email opens every transcript in the index — a list of readers we can write to, not a guest book.