Meta Platforms (META) is winning back investor confidence after a series of significant artificial intelligence developments this week rattled then reassured Wall Street.
Shares of Meta climbed 4.7% on Thursday after the company unveiled Muse Spark 1.1, an agentic coding model the company says is comparable to leading benchmarks from OpenAI and Anthropic.
The launch signals Meta’s growing ambitions in enterprise AI coding tools, a market segment where rivals have already established notable footholds.
BNP Paribas analyst Nick Jones described Muse Spark 1.1 as Meta’s first pay-to-use model, meaning the company now has a direct path to monetizing its AI products for the first time.
Developers can access Muse Spark 1.1 through Meta Model API, a new cloud hosting platform the company launched alongside the model on Thursday.
Meta’s stock initially dropped Thursday morning after a Reuters report revealed plans to double its data-center compute capacity from 7 gigawatts in 2026 to 14 gigawatts in 2027, reigniting concerns about runaway capital expenditure.
Those fears faded as investors absorbed the broader strategic picture, including progress on custom silicon that analysts believe could sharply reduce infrastructure costs.
At the center of that cost story is a chip called “Iris,” with production scheduled to begin in September 2026, representing Meta’s first custom data center AI chip developed under its Meta Training and Inference Accelerators program, known as MTIA.
Deutsche Bank analyst Benjamin Black estimates Meta could achieve up to 35% lower data-center costs in 2027 by combining Iris with Nvidia chips across its infrastructure.
“MTIA progress should improve the cost curve over time, particularly for inference and core recommendation workloads,” Black wrote in a Thursday note.
Broadcom (AVGO) is assisting Meta in designing the Iris AI semiconductors, while Taiwan Semiconductor Manufacturing (TSM) is set to contract-manufacture the chips for Meta’s data centers.
Jefferies analyst Brent Thill noted in a Wednesday report that Meta is likely to keep ramping up AI spending, with the company sharing at a recent conference that its Prometheus data center in Ohio came online in July.
Meta has indicated it expects capacity and infrastructure constraints to intensify going forward as demand for AI computing resources continues to accelerate industry-wide.
The combination of a monetizable AI model, a credible custom chip roadmap, and expanding data center capacity appears to have shifted the narrative away from overspending fears toward long-term competitive positioning.
Meta did not immediately respond to a request for comment.
