Learning Paths

Subject: Forget the Hype: This is How to Actually Analyze AI Investments

Friends,

AI investment analysis isn't about chasing the shiniest new transformer model. It's about understanding the underlying data infrastructure and the human capital that makes AI actually work. Everyone's drooling over the next OpenAI, but they're missing the forest for the trees.

Let's be brutally honest: most AI investment analysis focuses on the output (the model, the application) and completely ignores the input (the data, the talent). This is like judging a restaurant solely on the plating of the food, ignoring the quality of the ingredients and the skill of the chef.

Here's the breakdown of what actually matters, based on the 1,800 articles I've sifted through this week:

1. Data Supply Chain Mastery:

Forget fancy algorithms. The real gold rush is in data. Specifically:

• Data Acquisition & Curation: Can the company reliably and ethically acquire high-quality data? Are they building proprietary datasets? Think Scale AI, but applied to niche industries.
• Data Engineering Pipelines: How efficient and scalable are their data pipelines? Can they handle the volume, velocity, and variety of data required for training and inference? Look at companies like Databricks and dbt Labs – they're the unsung heroes.
• Data Governance & Security: Are they addressing data privacy, bias, and compliance issues proactively? A single data breach or biased model can sink an entire company. This is where companies like Immuta become critical.

2. Talent, Talent, Talent:

AI is a people-driven business. It's not about buying the best software; it's about hiring the best people who can use the software effectively.

Deep Domain Expertise: Does the team understand the specific* industry they're targeting? A general AI team trying to disrupt healthcare is a recipe for disaster. Look for teams with strong medical backgrounds paired with AI expertise.

• Data Science & Engineering Prowess: Can they build, deploy, and maintain complex AI systems? Are they attracting and retaining top talent? This is where you need to dig into the team's background, publications, and open-source contributions.
• Ethical AI Leadership: Is there a commitment to responsible AI development at the highest levels of the organization? This is non-negotiable.

3. Infrastructure & Compute:

While the first two are more important, you can't ignore the underlying infrastructure.

• Cloud Providers: Are they leveraging the right cloud infrastructure for their needs? AWS, Azure, and GCP all have their strengths and weaknesses.
• Specialized Hardware: Are they taking advantage of specialized hardware like GPUs and TPUs to accelerate training and inference? NVIDIA is the obvious player here, but keep an eye on emerging players like Cerebras Systems.

The Noise vs. The Signal:

| Metric | Noise

Subject: Forget the Hype: This is How to Actually Analyze AI Investments (Part 2)

Friends,

AI investment analysis isn't about chasing the shiniest new transformer model. It's about understanding the underlying economics. And that means building specific capabilities.

But let's be honest: most "AI tools" are just dressed-up APIs. You're not buying a solution; you're buying a dependency. So, what actually helps you build the skills to analyze AI investments effectively?

Forget the "AI-powered" dashboards. Focus on these three pillars:

1. Cost Modeling Prowess: Can you accurately predict the true cost of running an AI model at scale? This isn't just about GPU hours. It's about data wrangling, infrastructure, talent, and the hidden costs of model drift. Tools? Forget the fancy UIs. Master your Python and SQL. Learn to leverage cloud cost explorers from AWS, Azure, and GCP. Bonus points for understanding the nuances of spot instances and reserved capacity.
2. Scenario Planning Muscle: AI investments are inherently uncertain. You need to be able to model different futures. What happens if compute costs double? What if a competitor releases a breakthrough model? Tools like Monte Carlo simulators (easily built in Python) are your friend. Stop relying on static spreadsheets. Think dynamically.
3. Value Chain Decomposition Acumen: Can you break down the AI value chain and identify where the real value is accruing? Is it in the foundational models (Anthropic, OpenAI)? The infrastructure providers (Nvidia, CoreWeave)? Or the application layer (C3.ai, UiPath)? This requires deep domain expertise and a healthy dose of skepticism.

Here's my hot take: The biggest opportunity isn't in building more AI models. It's in building better tools to understand the economics of AI. Think about it:

This week, I'm watching:

• Anthropic: They're raising billions, but can they translate that into sustainable profitability? Their focus on customizable AI solutions and enhanced security features is smart, but the execution is everything.
• Nvidia: Everyone's darling, but their valuation is pricing in perfection. Keep a close eye on their gross margins. Any sign of compression could signal trouble.

My advice? Stop chasing the hype. Start building the skills you need to understand the economics of AI. Your portfolio will thank you.

Until next week,

Yves Mulkers
The Ins7ghts Newsletter

Subject: Forget the Hype: This is How to Actually Analyze AI Investments (Part 3) - The Untapped Goldmine

Friends,

Want to know the real AI gold rush? It's not in the models themselves, but in educating the masses (and the elites) on how to use them effectively. Everyone's so focused on building the next GPT-n, they're missing the massive opportunity to train the people who will actually deploy and manage these systems.

Think about it: every company is scrambling to "become an AI company." But how many actually have the internal talent to do so? Zero. Point. Nada.

This creates a massive market for AI education, far beyond your typical Udemy course. We're talking about:

• Executive Education: Training CEOs and boards on AI strategy, not just buzzwords.
• Upskilling Programs: Retooling entire departments (marketing, finance, HR) to leverage AI tools.
• Specialized Bootcamps: Creating a new generation of AI-savvy data analysts, product managers, and engineers.

The Players (and the Gaps)

My Hot Take: The biggest gap is in applied AI education for business users. Everyone's learning how to build a model, but nobody's learning how to use it to improve sales forecasting, optimize marketing campaigns, or automate customer service. The opportunity is in bridging the gap between AI theory and real-world business problems.

Specific Advice:

1. Look beyond the code: Focus on courses and programs that teach AI strategy, business applications, and ethical considerations.
2. Invest in specialized training: General-purpose AI courses are a waste of time. Find programs tailored to your specific industry or role.
3. Demand practical experience: Look for courses that include hands-on projects, case studies, and real-world datasets.
4. Don't underestimate the human element: AI is a tool, not a replacement for human intelligence. Invest in training that emphasizes critical thinking, communication, and collaboration.

The AI revolution won't be won by the best models, but by the people who know how to use them. Bet on education, and you'll be betting on the future.

Yves