Why did Meta stock jump 15% in one week?

Meta surged approximately 15% after launching Llama (Metapark 1.1), a commercial AI model priced far below competitors — $1.25 per million input tokens versus ~$5.52 for Claude and similar OpenAI pricing. The dramatic price advantage signals Meta is ready to compete directly in the AI cloud services market. Bank of America's simultaneous $835 price target upgrade amplified the move.

What is Bank of America's thesis for Meta at $835?

Bank of America believes Meta is evolving into a full AI infrastructure company — developing custom silicon chips, scaling to 14 gigawatts of compute, and selling AI services externally. This positions Meta similarly to Amazon (AWS) or Google (TPU/Cloud), where proprietary infrastructure creates a durable revenue moat. BofA sees 32% upside from current levels to reach $835.

Is Meta expensive at current prices after the 15% rally?

By most fundamental metrics, Meta is at or below its historical average valuation. The trailing P/E sits below the 5-year and 10-year averages. Price-to-Sales of 6.4x is also below the five-year average. The only elevated metric is Price-to-Free-Cash-Flow, which reflects the company's heavy capital expenditure cycle. On balance, analysts view the current price as fair-to-cheap for a company of Meta's quality.

Meta Surges 15%: Bank of America Sets $835 Target as Llama AI Ignites the Rally

By Yogurt · 2026-07-12 · Market Analysis

Bank of America upgraded Meta to an $835 price target — 32% upside from current levels. The catalyst is a cheaper-than-ever Llama AI model that positions Meta as the next AWS. Here's the bull case, the risk, and the one chart level that matters.

Bank of America's $835 Upgrade: Meta Is Becoming the Next AWS

On July 10, 2026, Bank of America elevated its Meta Platforms price target to $835 per share — a Buy rating implying 32% upside from current levels. The thesis is bold: Meta is quietly transforming from a social-media advertising machine into a full-blown AI infrastructure company, on par with Google and Amazon.

The bank's core argument centers on Meta's decision to develop its own custom silicon chips — a move that mirrors Amazon's Trainium/Inferentia path and Google's TPU strategy. By controlling its own compute hardware, Meta can dramatically lower the cost of running AI workloads, widen its profit margins, and accumulate vast proprietary data-processing capabilities. Bank of America estimates Meta will reach 14 gigawatts of compute capacity as part of its buildout — and capital expenditures could exceed $100 billion in 2027.

That capex number has rattled some investors. Wall Street will demand to see a clear revenue path justifying the spend. If Meta can show it is selling compute power and AI services — rather than just using it internally for Instagram recommendations and smart glasses — the market will accept the bill. If not, expect a sharp reaction when 2027 guidance is issued.

The Llama Catalyst: When Cheap AI Wins

The more immediate catalyst for last week's ~15% surge in Meta's stock was the commercial launch of Llama (Metapark 1.1) — a large language model that Meta is now making available via API, with real pricing that undercuts every major competitor.

The numbers are striking. Meta charges $1.25 per million input tokens, compared to Claude (Anthropic) at $5.52 and OpenAI at comparable pricing. On output tokens, Meta prices at $0.15 versus roughly $5.50 for alternatives. That is not a marginal discount — it is a structural pricing disruption. Models at this quality tier (benchmarks comparable to Claude Opus and GPT-4-class) at a fraction of the cost mean one thing: enterprise adoption accelerates.

More companies building on cheaper AI means more volume running through Meta's infrastructure. More volume means more revenue data, better model fine-tuning, and a moat that widens with every API call. This is the classic Amazon AWS flywheel — and it is now Meta's playbook.

Fundamentals: Is Meta Actually Cheap?

The valuation picture is surprisingly modest for a company in the middle of an AI arms race. Meta's trailing P/E of approximately 20x sits below both its 10-year and 5-year historical averages — meaning the stock is cheaper on this metric than it has been for most of the past decade. The Price-to-Sales ratio of 6.4x is also below the five-year average, and EV/EBITDA is running around the long-run mean.

On each fundamental metric, buying Meta today means buying it at average or below-average historical pricing — not at a bubble premium. The one exception is the Price-to-Free-Cash-Flow ratio, where heavy capex spending has compressed free cash flow temporarily, making the stock look slightly elevated on that measure. That is a known and accepted trade-off during an infrastructure buildout cycle.

Analyst consensus is uniformly constructive: virtually every major firm covering Meta has a Buy or equivalent rating, with price targets clustered above current prices. The market is not being asked to take a leap of faith — it is being asked to trust a proven compounder deploying capital into its highest-conviction growth project.

The Technical Picture: Two Scenarios

Meta's chart tells a more nuanced story. The stock broke above its 150-day moving average last week — a technically significant level that has acted as resistance multiple times this year. But it did so with a gap up: a jump in price with no trading activity in between, leaving an unfilled void on the chart around the $633 level.

Gaps have a statistical tendency to close. That doesn't mean the stock reverses immediately, but it raises the probability of a pullback to fill the gap before the next sustained leg higher. At the same time, Meta remains trapped inside a pattern of descending trendlines — a series of lower highs that began near $790. To technically confirm a bullish breakout, the stock needs to close above $690–$691 with strong volume and a solid bullish candle. Until that level breaks, the descending trend structure remains intact.

Two scenarios are on the table. In scenario one, Meta corrects modestly, closes the gap near $633 (where buyers may reenter), and rebuilds momentum for a subsequent breakout. In scenario two, it powers through $690–$691 now, with enough buying volume to invalidate the descending trend. The second path is possible — but without that high-volume breakout candle, the chart calls for patience over conviction.

The Bottom Line

Meta is not the stock it was in its worst moments. It has a cheaper AI model than any major cloud provider, a path to custom silicon that could transform its cost structure, and fundamentals that are not expensive by any historical standard. Bank of America's $835 target is not a fantasy — it is the price that emerges if the capex bet pays off and the AI monetization story unfolds.

The question is timing. If you want in, the two levels to watch are $633 (potential gap-close entry) and $691 (breakout confirmation entry). Missing either and chasing intraday means buying into a 14% premium with an unclosed gap beneath you — that's not a trade, it's a hope. Let the chart give you a signal first.