SerpApi.Org
Alphabet's Q2 2026 earnings call is underway, and anyone glancing at tonight's headline earnings-per-share number will see a figure that looks nothing like what analysts predicted — and has little to do with how Google's actual business performed. Alphabet holds roughly a 14% stake in Anthropic, which saw its valuation triple from $380 billion to $965 billion during the second quarter after a $65 billion fundraising round that Anthropic confirmed in its official announcement. Bank of America estimates that markup alone generated approximately $80 billion in unrealized gains, which flows straight through Alphabet's income statement under mandatory accounting rules, according to the bank's pre-earnings analysis. The bank projects Alphabet's reported GAAP EPS may land near $8.38 — nearly triple the $2.87 operational consensus tracked by Alphastreet and other consensus aggregators. The smart money will ignore that number entirely and look at Google Cloud.
Alphabet's Anthropic Stake Produces $80 Billion in Paper Profit
Exchange rate as of July 22, 2026; conversions are approximate.
This is not a one-time quirk. Alphabet is required by FASB Accounting Standards Update 2016-01 — which took effect in 2018 — to mark its minority equity stakes in private companies to fair value at the end of every quarter and run any gain or loss directly through net income. When Anthropic's valuation is $380 billion, Alphabet's ~14% stake is worth roughly $53 billion. When that valuation reaches $965 billion, the stake is worth roughly $135 billion — and the $82 billion increase appears on the income statement as profit, without a single dollar changing hands.
This structure already played out in Q1 2026, when Alphabet reported net income of $62.6 billion and GAAP EPS of $5.11 — a figure that immediately made headlines as an 81% profit surge. Strip out the $37.7 billion in unrealized gains on non-marketable equity securities (primarily Anthropic), and adjusted EPS was $2.62 — a one-cent miss against the $2.63 that analysts had modeled. Multiple financial analysts have reported that nearly half of Alphabet's Q1 net income came from marking up the value of its equity stakes rather than from advertising, cloud, or any product it actually sold. The Q2 version of that effect is roughly twice as large because Anthropic's valuation more than doubled between the two reporting periods.
For a reader deciding whether Alphabet is growing its business or just marking up an investment portfolio, the rule of thumb is simple: the number worth watching is operating income, operating margin, and Google Cloud revenue — not reported EPS.
Google Cloud Revenue Is the Number That Decides the Stock
Google Cloud generated $20.0 billion in revenue during the first quarter of 2026, growing 63% year over year — the fastest growth rate among the three major cloud providers, ahead of Microsoft Azure's 40% and Amazon Web Services' 28%, according to IG International's Q2 earnings preview. That pace made Google Cloud, for the first time, the primary growth driver of the entire company.
For Q2, analysts project Cloud revenue of approximately $22.2 billion to $22.8 billion, implying growth of 64% to 67% year over year, per StreetAccount estimates reported by CNBC. If that figure holds, Google Cloud will have grown faster than Azure and faster than AWS for two consecutive quarters — a combination that would validate the company's claim to be the AI cloud of choice for enterprise customers.
The backstory to that claim is a specific technology: Google's Tensor Processing Units. Unlike AWS and Microsoft, which primarily rely on NVIDIA's GPUs for AI inference, Alphabet has built its compute stack around custom ASICs it designed entirely in-house, as detailed in Google Cloud's TPU documentation. TPUs use a "systolic array" architecture — data flows continuously through thousands of arithmetic logic units without needing to repeatedly access memory — which delivers significantly higher throughput per watt than NVIDIA GPUs for the matrix multiplication operations at the core of large language model training and inference. The tradeoff is specialization: TPUs require Google's XLA compiler and work best with JAX, TensorFlow, and certain PyTorch configurations; they are less suited to small batch sizes or mixed AI-and-graphics workloads where GPU flexibility matters more.
That specialization becomes an enormous cost advantage at scale. When CEO Sundar Pichai disclosed on the Q1 earnings call that Alphabet's AI models were processing more than 16 billion tokens per minute via direct API, that throughput ran on TPUs rather than on chips for which Alphabet paid NVIDIA's markup. One analysis found that for every NVIDIA GPU Google deploys internally, it deploys approximately nine custom TPUs — a 1:9 ratio that illustrates just how deeply the proprietary silicon advantage is embedded in Alphabet's infrastructure economics.
Now Alphabet is commercializing that advantage externally. The eighth generation of TPUs — TPU 8t for training and TPU 8i for inference — was unveiled in April 2026, with the TPU 8i's SparseCore-Collectives Acceleration Engine specifically designed to reduce on-chip latency for agentic AI workloads. Alphabet is selling TPU capacity directly to external customers including Anthropic, OpenAI, and Meta, and a Blackstone-backed neo-cloud venture is scheduled to rent TPU compute directly to AI labs and enterprises beginning in 2027, per IG International's analysis. Management has indicated that hardware sales will begin contributing to revenue later in 2026. On tonight's call, investors will listen for any evidence that external TPU revenue is materializing rather than remaining a cost advantage used entirely in-house.
What the $462 Billion Backlog Actually Means
Google Cloud entered Q2 2026 carrying $462.3 billion in contracted but not-yet-recognized revenue — nearly double the level from the previous quarter, as reported in Alphabet's Q1 2026 10-Q filing with the SEC. For context, the total Google Cloud operating division generated roughly $20 billion in a single quarter; its backlog therefore represents more than five years of current Cloud revenue sitting in already-signed contracts.
CFO Anat Ashkenazi told analysts on the Q1 call that demand is exceeding supply: "Our cloud revenue would have been higher if we were able to meet the demand." Alphabet projects that just over 50% of that backlog will convert to recognized revenue within 24 months. If that conversion holds, it points toward Google Cloud revenue roughly tripling from current levels by mid-2028.
Analysts at Visible Alpha noted that Q2 revenue expectations have risen to $117.2 billion from $113.0 billion in February, driven by ad business resilience and Cloud strength. Cloud margins are a parallel question: operating margin expanded from 17.8% in Q1 2025 to a record 32.9% in Q1 2026. Looking ahead to Q3 2026, analysts now expect Cloud operating profit margin to settle near 31.2%, down from a prior estimate of 35%, partly because of the Wiz cybersecurity acquisition being absorbed into the unit and the costs of rolling out eighth-generation TPU infrastructure.
The Wiz acquisition — which closed in 2025 for $32 billion — added cybersecurity workloads and a customer base that Alphabet has been integrating into Vertex AI and Cloud's enterprise security stack.
Will $190 Billion in Spending Pay Off?
The most uncomfortable number heading into tonight's call is not revenue or EPS — it is free cash flow. Alphabet raised its 2026 capital expenditure guidance in April to $180 billion to $190 billion, up from a prior range of $175 billion to $185 billion, with management indicating that 2027 spending will "significantly increase" beyond even this elevated level, per CNBC's coverage of the Q1 earnings call. That $190 billion upper bound is channeled into data center construction, server hardware, GPU and TPU procurement, and network infrastructure — physical assets that underpin both Gemini model training and Google Cloud's enterprise AI services, as detailed in TradingKey's earnings preview.
As a direct consequence, Q1 2026 free cash flow fell 47% year over year to $10.1 billion, even as operating income rose 30% to $39.7 billion. First-quarter capital expenditure was $35.7 billion — up 107% from $17.2 billion in Q1 2025. With approximately 60% of infrastructure spending allocated to servers and the remaining 40% to data centers and networking equipment, the spending is heavily front-loaded against revenue that will not fully appear until contracted workloads deploy across the backlog window.
An Epoch AI analysis published in June 2026 found that the five largest cloud builders — Alphabet, Amazon, Microsoft, Meta, and Oracle — are spending on AI infrastructure 70% faster than their cash earnings are growing, and that their aggregate free cash flow will collectively reach approximately zero at some point this summer. That arithmetic makes Alphabet's backlog conversion timeline not merely a growth metric but a solvency-adjacent question: can contracted Cloud revenue ramp fast enough to fund the 2027 capex cycle?
The bull case is grounded in historical precedent. Alphabet's Cloud business spent years consuming capital before generating significant profit; it began reporting Cloud operating income separately only in 2022, and Cloud margins have since expanded from single digits to 32.9%. Guggenheim analyst Michael Morris maintained a Buy rating and $450 price target ahead of the release, calling the current setup an "attractive entry for full-stack AI leader." At roughly 61 analysts recommending Buy and an average 12-month price target near $430, the Street's consensus is that the spending is generative rather than destructive.
The bear case is captured in one statistic: capex grew 107% in Q1 while Cloud revenue grew 63%. Spending is, for now, outpacing the business it is meant to support. Freedom Capital Markets Chief Market Strategist Jay Woods framed the threshold plainly: "Wall Street wants evidence that AI is driving growth rather than simply driving expenses."
Regulation Is Now Structural, Not Episodic
Heading into tonight's call, Alphabet is navigating two EU regulatory developments that have moved from headline risk to structural operating constraint.
On July 2, 2026, the EU Court of Justice upheld the €4.1 billion (approximately $4.68 billion) Android antitrust fine — ending an eight-year legal battle and establishing a precedent that will strengthen future EU cases against Google's search self-preferencing and Play Store anti-steering conduct. The fine itself is manageable for a company that generated $109.9 billion in Q1 revenue alone, but it signals that EU enforcement capacity is accelerating rather than plateauing.
More consequential for the operating business is the European Commission's Digital Markets Act decision issued on July 16, 2026, which requires Google to share anonymized search query data with rival AI developers by January 2027 and to open 11 Android features — including voice activation and system-level search integration — to competing AI assistants by July 2027. Noncompliance exposes Alphabet to fines of up to 10% of global annual turnover. At Alphabet's current revenue scale, that is a potential fine larger than the Android antitrust case.
The DMA timeline matters specifically because Alphabet has been completing Gemini's integration as the default AI assistant on Android — a process that now must be designed to accommodate third-party AI assistants on equal terms before the July 2027 deadline. The Gemini 3.5 Pro delay, reported in mid-July, adds pressure: CEO Pichai had signaled a June release, but engineers are still working on coding performance benchmarks, and some analysts have flagged the delay as a question about Alphabet's AI product execution timeline heading into tonight's call.
How to Read Alphabet's EPS Tonight Without Being Fooled
If Bank of America's estimate proves accurate, Alphabet will report GAAP EPS somewhere near $8.38 tonight — compared with a consensus operational estimate of approximately $2.87. The gap exists because FASB ASU 2016-01 requires Alphabet to mark its Anthropic stake to fair value at the end of every reporting quarter and run the change through net income, regardless of whether the company sold a single share. This is not a discretionary accounting choice. It is a mandatory rule, which means the pattern will repeat every quarter that Anthropic's valuation changes — up or down.
Tax law expert Robert Willens, quoted in Fortune's analysis of Alphabet's earnings quality, raised the specific concern that "they're able to control or influence the value of one of their own assets" — because Alphabet is simultaneously a major customer of Anthropic (purchasing compute capacity), an investor in Anthropic, and the provider of the TPU infrastructure that underpins much of Anthropic's model training. That structural relationship means Alphabet's transactions with Anthropic can influence the commercial metrics that determine Anthropic's valuation, which in turn determines the line item on Alphabet's income statement. One Seeking Alpha analysis found that approximately 74% of the year-over-year increase in Alphabet's Q1 pretax income came from below the operating line — primarily these equity markups — rather than from advertising, Cloud, or any other revenue-generating product.
The numbers to bring to the earnings call are these: operational EPS of approximately $2.87, Cloud revenue consensus of approximately $22.2 billion, Cloud operating margin expectation of approximately 31–33%, and free cash flow. If Cloud revenue comes in ahead of $22.2 billion and Cloud margins hold, Alphabet has demonstrated that its infrastructure spending is generating operating leverage. If Cloud margins compress sharply or revenue disappoints, the $8 headline EPS will not save the stock.
Can Google Cloud Sustain 64% Growth?
Google Cloud's competitive position in 2026 is specific: approximately 11–13% of global cloud infrastructure spending by market share, according to Synergy Research Group data, but the fastest growth rate among the three major hyperscalers for at least two consecutive quarters. It holds that position primarily through two differentiated capabilities: its TPU-based AI infrastructure, which eliminates the NVIDIA GPU markup for AI-native workloads, and the Vertex AI platform, which provides an integrated stack from model training to enterprise deployment.
The competitive risk is structural. Microsoft's Azure grows at 40% in part because of its exclusive partnership with OpenAI, which routes enterprise GPT workloads onto Azure infrastructure. AWS grows at 28% because of its breadth — more than 200 managed services and the largest partner ecosystem in the market. Google Cloud's AI advantage is real but narrower: it is best-in-class for JAX and TensorFlow workloads, and for customers already building on Gemini's model family, but it still trails AWS on service breadth and Azure on enterprise Microsoft 365 integration.
Meta's newly announced "Meta Compute" initiative signals that a fourth large-scale AI infrastructure competitor may emerge within the next two to three years — though it remains far from operational at hyperscaler scale today.
Waymo, Alphabet's autonomous vehicle subsidiary, now delivers approximately 500,000 paid robotaxi rides per week across more than ten US cities — a tenfold increase from 50,000 weekly rides as recently as May 2024, and a milestone that makes Waymo the operational leader in commercial autonomous vehicle service globally. It remains a rounding error in Alphabet's consolidated financials for now, but the robotaxi business has reached a scale where unit economics are becoming measurable rather than theoretical.
The earnings call is scheduled to run approximately one hour. What management says about Cloud margin trajectory, backlog conversion pace, TPU external revenue timing, and the Gemini 3.5 Pro development timeline will carry more information than any single reported number — including one that may print at $8.38 because of an accounting convention most readers will not understand until they look past the headline.
Frequently Asked Questions
Why does Alphabet's reported EPS look so much higher than what analysts predicted?
Alphabet is required by mandatory US accounting rules (FASB ASU 2016-01) to revalue its minority stake in Anthropic at market value at the end of every quarter and record any increase as income on its profit-and-loss statement — even when no cash changes hands. Anthropic's valuation rose from $380 billion to $965 billion during Q2 2026, following a $65 billion fundraising round. Alphabet holds roughly 14% of Anthropic, meaning that valuation jump produced approximately $80 billion in paper gains that flow directly through reported net income and inflate GAAP EPS. This has nothing to do with Google's ad revenue, Cloud subscriptions, or any product the company sold. Bank of America estimates reported GAAP EPS may reach approximately $8.38, compared with a consensus operational EPS of approximately $2.87. The gap between those two numbers is almost entirely the Anthropic accounting effect.
What is a Tensor Processing Unit and why does Google's TPU advantage matter for investors?
A TPU is an application-specific integrated circuit that Google designed from scratch to accelerate the matrix multiplication operations at the core of AI model training and inference. Unlike NVIDIA GPUs, which carry general-purpose graphics circuitry that is useful for gaming and rendering but irrelevant to AI, TPUs trade that flexibility for dramatically higher throughput per watt on AI workloads. Google uses a "systolic array" architecture — data flows in a continuous wave through thousands of arithmetic logic units, requiring far fewer memory read-write cycles than a GPU performing the same computation. The practical implication for investors is that Google does not pay NVIDIA's markup when running its own AI workloads. At a scale of 16 billion tokens per minute, that cost gap is substantial. Now that Alphabet is selling TPU access externally — to Anthropic, OpenAI, Meta, and through a forthcoming Blackstone-backed neo-cloud venture — the TPU is becoming a revenue source rather than only a cost advantage.
Why is Google Cloud growing faster than AWS or Microsoft Azure if it has less market share?
Market share and growth rate reflect different things. AWS leads with roughly 30% of global cloud infrastructure spending and the broadest service catalog in the industry. Azure follows at roughly 24% with the strongest Microsoft 365 and enterprise integration. Google Cloud holds approximately 11–13% of the market. But growth rate reflects where new enterprise workloads are being placed, and in 2026 that is disproportionately AI and machine learning. Google Cloud's TPU infrastructure and Gemini AI model integrations give it a specific advantage for AI-native enterprise workloads — the fastest-growing category of cloud spending. Analysts have noted that AI workloads now account for about 19% of all cloud spending, up from 8% in 2023, and those workloads disproportionately favor platforms with purpose-built AI silicon. That dynamic is what drives Google Cloud's 63–64% growth rate even from a smaller base.
What are the EU regulatory risks that investors should watch on tonight's call?
Two specific actions have escalated from risk to operating constraint. First, on July 2, 2026, the EU Court of Justice made permanent a €4.1 billion (approximately $4.68 billion) Android antitrust fine against Google, closing an eight-year legal case while establishing precedents that strengthen pending EU investigations into search self-preferencing and Play Store conduct. Second, on July 16, 2026, the European Commission issued binding Digital Markets Act orders requiring Google to share anonymized search data with rival AI developers by January 2027 and to open 11 Android features — including voice activation and system-level search integration — to competing AI assistants by July 2027. Noncompliance carries fines of up to 10% of global annual turnover. For a company generating roughly $450+ billion annually, that potential penalty is structurally significant. Management is likely to face direct questions on compliance costs, product redesign timelines in Europe, and whether the DMA obligations affect Gemini's Android rollout schedule.
Related Articles
