Meta is done playing catch-up quietly. The company unveiled Muse Spark, its first flagship large language model from the newly formed Superintelligence Labs, and paired it with an eye-watering spending plan that makes clear it intends to buy its way to the AI frontier if it has to.
What Muse Spark is
Muse Spark is the first flagship model built under Chief AI Officer Alexandr Wang’s Superintelligence Labs. Meta says it delivers competitive performance on multimodal perception, reasoning, health and agentic tasks — at a fraction of the compute cost of its older Llama 4 mid-size variant. Efficiency is the headline claim: comparable capability for less compute would let Meta serve AI across its billions of users without the costs spiraling out of control.
The spending shock
The number that grabbed Wall Street’s attention was the capital plan: AI capital expenditures of $115-135 billion for 2026, nearly double last year. Meta also raised its planned investment in an El Paso, Texas data center from $1.5 billion to over $10 billion, expanding it to one gigawatt of capacity by a 2028 launch. This is infrastructure spending on the scale of national energy projects.
Why Meta is going all in
The aggression reflects anxiety. Meta has trailed OpenAI and Google in the perception of who leads generative AI, and the expensive recruitment of Wang signaled a reset. Muse Spark plus record capex is the company betting that talent, compute and scale can close the gap — and that being behind in AI is a far costlier outcome than overspending to catch up.
The competitive context
Rivals are not standing still. Amazon launched its Quick AI work assistant and brought OpenAI models to Bedrock, while its custom-silicon business (Graviton, Trainium, Nitro) passed a $20 billion annual run rate. The hyperscalers are all pouring tens of billions into AI infrastructure at once — a capital arms race where the table stakes keep rising.
The investor’s dilemma
For shareholders, the bet is double-edged. Spending $130 billion to win AI could cement Meta’s future or dent returns if the payoff lags. The efficiency story behind Muse Spark is meant to reassure that the money buys durable advantage, not just bigger bills. Markets will watch whether usage and revenue justify the outlay.
The bottom line
With Muse Spark and a near-doubled capex plan, Meta has declared it will spend whatever it takes to compete at the AI frontier. The model suggests progress; the budget suggests urgency. Whether it closes the gap with OpenAI and Google — or just widens the industry’s spending spiral — is the trillion-dollar question.
Photo: public domain via rawpixel