When Alphabet reports third-quarter earnings on Oct. 28, the headline numbers may matter less than one specific question: can Google Cloud turn its expensive new AI computing capacity into actual sales of its in-house AI chips?
That's the view from Bank of America, which in a note Monday laid out its expectations for the quarter. The bank's analysts forecast Alphabet will post revenue of $111.4 billion and earnings per share of $3.07, both slightly ahead of the consensus on Wall Street, which sits at $111.0 billion and $2.98 per share.
The gap between those numbers and what investors focus on could be wide. BofA sees Google Cloud as the most likely place for a surprise, but it comes with a catch: building out AI infrastructure is expensive, and the costs show up before the revenue does.
Why AI capacity costs are the swing factor
Alphabet, like other tech giants, is spending heavily on data centers, servers, and specialized chips to support the boom in artificial intelligence. The company designs its own AI processors, called tensor processing units, or TPUs, which are used to train and run AI models on Google Cloud.
Adding all that capacity brings a wave of new costs. Depreciation — the accounting charge that spreads the cost of equipment over its useful life — rises as servers are installed. Power bills climb, and so do staffing costs to run and maintain the infrastructure.
The problem is that these costs ramp up before demand fully materializes. If more customers run AI workloads on Google's infrastructure, utilization rises, and Cloud can spread those fixed costs across more revenue, which supports profit margins. But if capacity grows faster than the workloads that use it, underused servers can squeeze margins even if revenue holds up.
That's why TPU sales matter. They're a signal of whether customers are actually adopting Google's AI infrastructure, not just whether the company is building it.
What else to watch in the report
BofA also noted that Meta's third-quarter advertising performance could influence expectations for Alphabet's Search business. Since both companies rely heavily on digital ads, a strong or weak showing from Meta can color how investors view Alphabet's ad revenue.
But the bigger discussion is likely to be forward-looking. Investors will want to hear what management says about the pace of AI capacity spending, how quickly it expects to monetize that investment, and what it means for next year's operating margin and free cash flow — the cash left over after expenses and capital investment.
Alphabet has been talking up its "agentic" AI plans, which refer to AI systems that can take actions on their own, like booking appointments or managing tasks. The timing of those products and how they might drive future revenue will probably be a key topic on the earnings call.
What it means for investors
For everyday investors, the takeaway is that Alphabet's AI buildout is largely a fixed-cost bet. Once the company spends on data centers, chips, and operations, the near-term swing is how much of that capacity gets used.
Strong TPU-driven demand would be a positive sign because it implies higher utilization and a better chance of holding Cloud margins up while spending rises. That would support the case that Alphabet's heavy investment is paying off.
The risk is timing. If costs climb ahead of revenue, investors may look past a modest beat or miss on the quarter's headline numbers and focus on what it means for next year's profitability. In that scenario, even a solid quarter could be overshadowed by concerns about margin pressure.
So the market's reaction on Oct. 28 may hinge less on whether Alphabet beats or misses by a few cents, and more on how management frames the AI capacity expenses and the pace of monetization. That's the real test.
For context, the broader tech sector has been sensitive to AI spending narratives. Recent earnings from other big tech names have shown that investors are watching closely how companies balance AI investment with profit growth. The wobble in AI stocks ahead of this earnings season shows how quickly sentiment can shift.
Alphabet's cloud business has also faced operational challenges. A recent outage in London disrupted UK banking apps, a reminder that reliability is critical for cloud customers. While that incident is unlikely to have a major impact on the quarter, it underscores the operational risks that come with scaling infrastructure.
On the positive side, Alphabet's cloud unit has been gaining traction with enterprise customers. A partnership with Cognizant to scale enterprise AI using Google's Gemini models is one example of how the company is trying to expand its cloud footprint.
Ultimately, BofA's $111.4 billion revenue call makes Google Cloud's cost story the real test on Oct. 28. Investors will be listening closely for any clues about whether the AI bet is starting to pay off.
