AI infrastructure
The physical and software stack — chips, clusters, data centres, networking, and orchestration — that training and serving large models actually runs on.
Learn first
AI infrastructure is the stack beneath the model: the chips (GPUs, TPUs, custom ASICs), the high-speed interconnects that link thousands of them into a cluster, the data centres built around their power and cooling demands, and the software that schedules jobs across all of it. Without it, a training run stays a spec sheet. The hard part is that each layer — silicon, fabric, facility, orchestration — requires years of lead time and specialised capital, so constraints here are felt long before they show up in benchmark results.
In megatrends
Related players
Finds citing this concept
Overall = breadth + depth + substance + ½·novelty + freshness (a small boost that halves each week)
Anthropic's $15B compute bill
SpaceX's IPO filing is the first audited look at what a frontier lab pays to run: Anthropic owes $1.25 billion a month through 2029 — to a Musk company training its rival.
Google forks the TPU
For the first time in a decade of TPUs, Google built two chips instead of one — a training chip and a serving chip — because the economics of the two jobs have pulled apart.
AI chips get an A-rating
CoreWeave borrowed $8.5 billion against its GPU clusters at a rate normally reserved for utilities and railroads — the first time graphics chips have been rated safe enough for pension funds to lend against.
The paper that spooked the chip market
A Google Research paper on squeezing an AI model's working memory by roughly six times spooked the memory-chip market — because cheaper inference could mean fewer of the expensive chips that serve it.
The penny-warrant template
Meta committed to roughly $100 billion of AMD chips and handed AMD a warrant for 160 million shares at a penny each — the same terms OpenAI signed five months earlier, now run a second time.
How this connects
Tap a node to open it