Meta (META): The Cheapest Mega-Cap Tech Bet
Meta (META) has become one of the strongest buy ideas because the upside here is larger than in peers that already ran up. A 12-month price target of $750 sits on a strong buy rating. Both Microsoft (MSFT) and Meta (META) spent months "in the penalty box," but Microsoft (MSFT) has started to move the right way, while Meta (META) still holds more room to rise.
The core ad business stays very healthy. Growth topped 30% in the first half of the year. That pace will slow against tougher year-over-year comparisons, dropping to low-to-mid 20s over the next few quarters, but the business is strong. On valuation the stock trades at only about 13 to 14 times the calendar 2028 estimate - by far the cheapest of the mega-cap tech names.
Two big worries hang over the stock: whether Meta (META) can turn AI into money, and regulatory uncertainty. Both look largely priced in already.
Non-Advertising Upside Not Yet in Estimates
Meta (META) already turns AI into ad revenue very well. The next leg is non-advertising income, and most of it sits outside current estimates:
- Subscription-type products.
- API revenue from its models.
- Possibly selling access to compute as an AI offering.
Non-ad "other revenue" was about $1 billion this quarter, roughly 2% of sales - still small, so there is a lot of room to grow. The plan: put several open-source models on a platform, then charge the enterprise market API fees on them. These would be priced very low. Leading closed frontier models keep their own place in the market, but there is real demand for open-source models that are not China-based, and that gap is the opening for Meta (META) and Microsoft (MSFT).
From those models Meta (META) can also build new tools - business AI agents and other offerings - inside its own products. This is low-hanging fruit: build momentum and show revenue over the next few quarters. There is even talk of a cloud infrastructure business selling compute to outside customers, which would compete with AWS, Microsoft Azure (MSFT), and Google (GOOGL).
AI Compute Feeds the Ad Machine
Meta (META) is an ads-based business and has always used AI compute to power ads, and it does that well. So the heavy capex is not wasted - those GPUs go straight to work driving the business. In a world of large language models, Meta (META) needs to own its own frontier model, because that shapes how it trains its ad models. Controlling its own destiny there is reason enough to train its own frontier models.
Investors struggle with the time it takes to earn back these investments. The payback runs through the core business: better ads, or new places to show ads. Extra upside could come from renting out GPUs or selling access to the frontier models, but those are bonus paths, not the plan.
Comparing Meta (META) to model labs like OpenAI and Anthropic, or to the hyperscalers that spend heavily on compute, is not apples-to-apples. Meta (META) has a strong core business, so it already turns AI into money better than OpenAI or Anthropic can, and it earns much fatter margins than a cloud provider (CSP) renting out chips. Zuckerberg says everything is on the table and has floated renting compute out, but the company clearly knows how to make money from this. The ask to investors: trust Zuckerberg and be patient.
AI Ads for Small Businesses
Meta (META) already does this and should tell the story more. As an ads business, it makes money when advertisers create ads that convert into customers. A small business often lacks the time or skill to make good ads or short video clips, so it either skips Facebook (META) ads or runs weak ones that convert poorly.
Anything that makes ad creation easier grows the pool of advertisers. If a small business can use generative AI to make a good ad or even a video, more small businesses will advertise. Meta (META) controls the quality, so it can make sure those AI-made ads convert. That widens Meta's (META) footprint, lifts conversions, and lets Meta (META) charge more. This sits right inside the core business and is not far-off science fiction. The clear message to investors: we buy GPUs, we train LLMs, and those LLMs grow the core business.
Near-Term Risk
Free cash flow is near zero and the share buyback program is largely paused. That leaves the stock exposed in the near term to sentiment - how people feel about AI spending and payback timelines - which could bring volatility.
