Alphabet is routinely discussed as a collection of parts — a search business under regulatory pressure, a cloud business chasing Microsoft and Amazon, and an AI lab racing OpenAI and Anthropic. That framing misses the mechanism that actually matters to GZC's thesis: these are not three separate bets. They are one flywheel, and the connective tissue is capital.
Search still throws off enormous free cash flow. That cash is what allows Alphabet to fund a data center and custom-silicon buildout at a scale few competitors can match without straining their balance sheets. The buildout, in turn, produces the compute that both defends Search against AI-native challengers and gives Cloud a differentiated product to sell. Each piece reinforces the other two. Valuing Alphabet as "Search plus some optionality" misses where the durable advantage actually sits.
Search Funds the Buildout, the Buildout Defends Search
The bear case on Google Search has been the same for two years: generative AI assistants and answer engines erode the query-and-click model that built the company. The first-quarter 2026 print is the clearest rebuttal to date. Alphabet reported revenue of roughly $110 billion for the quarter, and management characterized Search advertising growth as continuing above the mid-teens rate, with AI Overviews increasingly treated as a monetization layer rather than a cannibalization risk. Search is not a legacy business in decline — it is a still-growing cash engine that happens to be getting an AI-native upgrade at the same time.
That cash engine is precisely what lets Alphabet guide to $175–185 billion of 2026 capital expenditure without the market treating it as existential risk to the balance sheet the way it might for a less cash-rich competitor. This is the part of the ecosystem argument that gets lost when analysts model Search and Cloud as separate businesses: Search's durability is the reason Alphabet can out-invest rivals in the infrastructure race, and that infrastructure is increasingly what keeps Search itself competitive.
TPUs: The Silicon Advantage Nvidia Can't Undercut
At Google Cloud Next in late April, Alphabet formally introduced its eighth-generation Tensor Processing Units, split for the first time into a dedicated training chip and a dedicated inference chip. This is a meaningfully different position than nearly every other hyperscaler occupies. Amazon and Microsoft still rely overwhelmingly on external GPU supply for their AI infrastructure. Alphabet designs its own accelerators end to end, which changes the economics of every dollar of capex it spends: a portion of that spend builds a proprietary asset rather than simply renting compute at Nvidia's margin structure.
This is also where the ecosystem argument connects to GZC's broader BTT framework. The custom-silicon buildout is not a side project — it now touches Alphabet's own Cloud economics, its relationship with external silicon customers, and increasingly the AI stacks of companies outside its own ecosystem (Google's TPU program has become a customer of the same custom-chip supply chain GZC tracks elsewhere in this coverage universe, including Broadcom's ASIC business).
Cloud's Inflection and the Backlog Nobody Is Pricing In
Google Cloud grew revenue 63% year over year in the first quarter of 2026 — the first time in several years that Google Cloud's growth rate has outpaced Microsoft Azure's. More important than the quarterly growth number is the backlog: Alphabet closed 2025 with a Cloud backlog north of $240 billion, more than double the prior year, an unusually explicit form of forward revenue visibility for a business the market often still treats as a distant third to AWS and Azure.
Free cash flow margin compressed sharply as this capex ramped, and that compression is the most common reason cited for the market's muted reaction to otherwise strong quarters. GZC views this the way it views forced capex anywhere else in the BTT coverage universe: near-term margin pressure from building an asset base is not the same as a deteriorating business, and depreciation headwinds are a temporary drag against a backlog that is growing faster than the spend required to service it.
GZC Thesis Summary
We do not think of Alphabet as a Search stock, a Cloud stock, or an AI-silicon stock — we think of it as the clearest public example of a company where each of those businesses is capitalizing the others. Search's cash generation funds a custom-silicon and data-center buildout that few competitors can match without external financing; that buildout defends Search's competitive position while making Cloud a differentiated seller of proprietary compute. The market's tendency to value each segment in isolation, discounting Search as legacy and Cloud as sub-scale, is in our view the core mispricing in the name. We track GOOGL as a high-conviction position in the Technology pool on this basis, and will publish follow-on research addressing the company's competitive moat and our specific bull-case entry framework.
