Brian Nowak has covered Meta Platforms since it was still called Facebook. On June 6, the Morgan Stanley analyst went back to his clients with a message he has delivered before, and which the market keeps declining to accept: the stock is undervalued. Investor sentiment toward Meta, he wrote, lags every other trillion-dollar technology company, and the reason is easy to identify. It is the number $380 billion.
That is the size of Meta’s capital-expenditure plan, and the market treats it as a black box. Wall Street can see the spending; it cannot see what the spending buys. Nowak’s argument is that the market is reading the capex as a cost to be recovered when it should be reading it as a moat being built, an infrastructure advantage that compounds for years and that competitors cannot match.
The logic runs through everything Meta is building. The data centers, chips and models funded by the $380 billion are the foundation of the company’s AI products, from the models behind its recommendation systems to the new AI search feature it launched on Facebook this week. Each dollar of spending lowers the cost of serving ads and improves the quality of targeting, and the combination of Meta’s data with its own compute is an advantage that no rival can buy off the shelf.
The timing of the analyst’s call and the product launch is convenient for his case. AI Mode, which lets Facebook users ask questions and receive summarized answers drawn from the platform’s posts, is the first product that tries to turn Meta’s AI spending into revenue, and analysts have projected it could add $10 billion a year. In Nowak’s telling, the launch is evidence that the spending is producing products, not just costs.
The bear case is equally clear, and it has kept Meta’s valuation below its peers for a reason. A $380 billion spending plan with no guaranteed return is a large bet on a single assumption, that AI will transform advertising fast enough to pay for the infrastructure before it depreciates. The market remembers the metaverse, when Meta spent tens of billions on a vision that took years to produce anything usable, and it is not eager to relive that experience.
There is also the question of how the spending is managed. Meta’s ad business is a cash machine, but even a cash machine has limits, and the company is borrowing and issuing debt to help fund the buildout. If AI revenue arrives slower than planned, the spending will compress margins, and the stock will be punished for the very ambition that Nowak is praising.
The comparison to Meta’s peers cuts both ways. Microsoft, Google and Amazon are spending comparable sums on AI infrastructure, and their stocks have not been punished for it, because their spending is attached to businesses that already generate huge cloud revenue. Meta’s spending is attached to advertising, a business that the market assumes AI will disrupt before it enhances. Nowak’s counter is that Meta’s data advantage makes its ad business the most defensible of all, and that the market is discounting an asset it cannot see.
What would change sentiment is evidence, and the next few quarters will supply it. If AI Mode shows up in ad pricing, if the recommendation upgrades lift engagement, if the capex guidance comes down without growth slowing, the black box opens. Nowak is betting that the evidence arrives, and that the market’s skepticism about $380 billion turns out to be the entry price.
Meta’s financial position gives the spending room to run. The ad business has been generating free cash flow in the tens of billions each year, and the company has said it can fund the AI buildout while continuing to return capital to shareholders. The market’s concern was never that Meta could not afford the plan; it was that the returns on the spending would take too long to arrive, and Nowak’s argument is that the timeline is shorter than the market assumes.
The analyst’s track record gives his call weight. Nowak has covered Meta and its predecessors for more than a decade and has been consistently early on the company’s ad products, and his calls have coincided with some of the stock’s strongest runs. The market has also been wrong about Meta before: the stock was written off after the Cambridge Analytica scandal and after the metaverse pivot, and it recovered both times when the ad machine reasserted itself.
The call is ultimately a bet on management. Zuckerberg has a history of making large bets that look reckless until they work, and the market has oscillated between rewarding and punishing him for it. Nowak’s position is that the current punishment is overdone, that Meta is building the infrastructure of the next decade at a price the market has not yet accepted. If he is right, the stock is cheap. If he is wrong, the $380 billion will be remembered the way the metaverse spending was, as the price of a vision that did not pay.
