Meta stock closed at $613.62 on Sept. 3 and finished the day up about 4%. The broader market was up roughly 1%. Something was moving Meta specifically.
Bank of America analyst Justin Post published a note, shared with TheStreet, laying out why he thinks the stock has more room to run. He kept his Buy rating and his $810 price target. That target implies about 32% upside from where the stock closed.
What BofA says about Meta’s new AI model
Meta released Muse Spark 1.3 on Sept. 2. It is an updated model built for coding and agentic tasks. It is available through Muse Code and the Meta Model API.
The bank noted that Meta released this model just about a month after Muse Spark 1.2. That cadence is worth paying attention to. A month between frontier model updates is fast by any standard.
Muse Spark 1.3 uses roughly 20% fewer tool calls and about 25% fewer tokens than 1.2 to finish comparable engineering tasks, VentureBeat reported.
The model kept the same pricing that shipped with 1.2, at $1.25 per million input tokens and $4.25 per million output tokens. Meta is competing directly with Anthropic’s Claude Code and OpenAI’s coding tools in the agentic coding category.
More Meta:
Meta sent a warning to its glasses pranksters
Meta layoffs take disturbing turn in new lawsuit
Meta business model in trouble from $1.4 trillion lawsuit
The new model handles longer-running tasks better than its predecessor. It can manage multiple workflows in a single conversation. It pulls context from different sources and follows complex instructions more reliably.
BofA sees the agentic improvements as especially important because of what Meta is building next. Meta is working on a consumer AI agent internally codenamed Hatch, first reported by The Information.
Meta has not confirmed a commercial name or launch date. A model that handles longer-horizon agentic tasks better is a more capable foundation for a product like Hatch.
Why Meta’s custom chip strategy matters for investors
There is a second piece to the note, and it’s about chips.
Broadcom said on its second-quarter 2026 earnings call that it expects to deliver three generations of Meta’s custom Training and Inference Accelerator, or MTIA, through 2027, CNBC reported.
It also said it has visibility into roughly three gigawatts of Meta deployments through 2028. Production shipments are expected to start in the fourth quarter of 2026.
This sits within a larger Broadcom-Meta partnership announced in April 2026 and running through 2029. The initial commitment was over one gigawatt.
BofA estimates those deployments could represent 15% to 20% of Meta’s total AI capacity. Alphabet made the same bet with its TPUs years ago, and it paid off.
Companies that build their own silicon tend to get structural advantages in cost and performance over time as their AI workloads scale, and Meta is following the same playbook.
BofA sees the agentic improvements as especially important because of what Meta is building next.Bloomberg / Getty Images
Where Meta stock stands on valuation
At around $617 at the time of the note, Meta was trading at about 18 times what analysts expect it to earn in 2027 on a GAAP basis. That is roughly 15 times 2028 estimates.
Historically, Meta has traded at about 21 times earnings. The S&P 500 right now is at about 20 times. Meta is trading below both its own history and the index it is a part of.
BofA’s $810 price target is built on 24 times 2027 GAAP EPS. The bank argues that premium to the market is reasonable, given Meta’s growth rate and the size of the AI opportunity still ahead of it.
Regulatory pressure is another variable. Meta has been under significant legal scrutiny. Any meaningful reduction in that overhang would give the stock additional room to move, BofA argues.
The 52-week range for Meta stock runs from $520.26 to $790.80. The stock is sitting in the lower half of that range even after the Sept. 3 gains. BofA sees the current price as a window before execution confidence starts pushing the multiple toward its historical average.
BofA flags these 5 risks for Meta investors
BofA lists five specific risks in the note.
Advertising business: Most of Meta’s revenue comes from digital ads. Ads are sensitive to the economy. A slowdown hurts Meta disproportionately.
Spending: Meta is pouring money into AI. BofA expects that to weigh on margins.
The fixed cost base: The more Meta builds out physical infrastructure, the less flexibility it has if business turns.
Competition: AI-native platforms could pull users and ad dollars away from Meta’s apps.
Regulation: Teen-safety cases and ongoing litigation could produce outcomes that are hard to plan around.
Despite those risks, BofA’s central argument is that the stock is not pricing in enough of the AI upside. At 18 times 2027 earnings, Meta is cheaper than it normally trades and cheaper than the market. That is the entry point the note is making a case for.
The next things to watch are whether Hatch launches on the timeline The Information reported, whether Muse Spark 1.3 adoption picks up through Muse Code and the Meta Model API, and whether MTIA production shipments begin as scheduled in the fourth quarter.
Each of those is a data point that either builds or undercuts the bank’s thesis.
If the AI execution holds, the bank sees a path to $810. If it does not, the risks it named are real. The stock will feel them either way.
Related: Mark Zuckerberg sends shocking message to Meta employees