Background: Every major tech company is pouring billions into AI, and behind every model that can write, code or analyse is an extremely powerful chip. In fact, Meta has already spent an eye-watering $72 billion this year alone on Nvidia chips to fuel its AI ambitions.

What happened: Meta is now in talks to spend billions more, but this time on Google’s AI chips, known as TPUs, for its data centres starting in 2027. It’s a surprising turn because Google has historically kept TPUs locked inside its own data centres, never selling them externally.

What else: If the deal goes through, it marks a major strategic shift for Google and could help it claw away as much as 10% of Nvidia’s annual revenue. And for Meta, it’s a move toward diversifying chip suppliers instead of relying almost entirely on Nvidia.

What's the key learning?

💡Nvidia might be the king of the AI chip world right now, but the kingdom is beginning to feel a little crowded.

💡Nvidia currently controls 85 - 90% of the semiconductor market, but big tech companies like Google, Amazon and Meta are increasingly building their own AI chips to reduce dependence on Nvidia’s pricing and supply.

💡These same tech giants also account for more than 50% of Nvidia’s revenue, meaning Nvidia’s biggest customers are quickly becoming its most serious competitors as they race to develop in-house alternatives.