AI Strategy Case Studies for Product Managers
A must Read for Everyone
1. Why Meta is so bullish about Glasses? Is this a new Platform Shift?
Glasses are positioned as the ideal form factor for personal superintelligence because they are the only device that can simultaneously: see what the user sees, hear what they hear, talk to them throughout the day, and generate a UI in their vision in real time.
Regarding the platform shift, Glasses may not replace the Phones rather reduce the reliance on them for simpler task ( like reading messages or start Navigation etc ) and make users more immersed with the surrounding.
The long-term goal is an always-on experience (which the user controls) where the system can think about the context of conversations and generate a UI from scratch when an application is needed.
The only thing which is to watch out for is how quickly people will want to adopt this. Although Meta has made a good progress on the Design of the glasses like making them aesthetically good looking.
2. Why is it so hard to compete with Nvidia?
It’s not just the chips.
Competitors are trying to create a better chip but Nvidia is trying to create an ecosystem which locks in the customer.
1) The Software Lock-In (CUDA): For over a decade, they’ve built a software ecosystem that is now the industry standard. Migrating away from CUDA is so difficult and expensive that most companies don’t even try. It’s the ultimate switching cost.
2) The System Lock-In (NVLink): They aren’t just selling chips anymore. They’re selling a fully integrated, rack-scale system where the CPU, GPU, and networking are optimized to work together. This tight integration delivers performance that other systems can’t match.
In a power-constrained world, the most critical metric is performance-per-watt.
3. Why OpenAI is teaming up with Broadcom to make Custom Chips?
Why can’t they simply use GPUs?
Before that let’s understand the difference between GPUs and ASICs —
GPU — This chip is very flexible and programmable. It is made of many cores which are working parallel. Since it can be used for tasks that haven’t even been invented yet, it’s great for general AI development.
ASIC — ASIC is a custom-built for only one specific job. Because it is designed perfectly for that one job, it is usually much faster at it and cheaper to make. It is not very flexible.
Open AI needs something in between which is customisable but less flexible. But Why ?
Because ChatGPT, is used by a massive number of people. This means they know exactly what kind of tasks their chips are performing every single second.
let’s take an example, one of the task is “Reasoning” but it requires a lot amount of time/Compute which is why it is slightly slow. OpenAI might build custom Chips that is “tuned precisely for reasoning inference”.
The goal is to improve reasoning so much that it brings down costs and increases usage.
So OpenAI will still need GPUs and NVIDIA continues to providing but in addition to that for the very specific task, they might think about Broadcom
4. The AI Video Race: Why Google Has a Huge Advantage?
Making AI videos takes an incredible amount of computer power. This raises the question of which company is best positioned to succeed in this new field.
Creating a 10-second AI video requires a massive amount of data, which needs very powerful computer chips (GPUs).
OpenAI has the hottest technology, but they don’t have enough GPUs to let everyone use it. That’s why they’re releasing it slowly. They are constrained by compute.
Google’s Big Advantages:
1) Infrastructure: Google already has a global network of massive data centres.
2) Custom Chips (TPUs): Google designs its own AI chips called TPUs (Tensor Processing Units), which are specifically made for its AI software. They don’t have to rely as much on buying chips from Nvidia.
3) Distribution (YouTube): They own YouTube, the world’s biggest video platform, which is the perfect place to release AI video tools.
4) Business Model (Ads): They can offer AI video tools for free and make money from ads, while others might have to charge users directly.
For Full Detailed cases Studies and AI Strategy — Download this Book ( 5/5 Rated )
Resources
About Author
Shailesh Sharma! I help PMs and business leaders excel in Product, Strategy, and AI using First Principles Thinking. For more, check out my Live cohort course, PM Interview Mastery Course, Cracking Strategy, and other Resources
 | A guest post by
|
Thanks for writing this, it clarifies a lot, and your analysis of Meta's potential platform shift with glasses and Nividia's ecosystem lock-in really builds on your earlier pieces about AI strategy, making me wonder how the 'always-on' personal superintelligence will shape our daily lives and digital autonomy.