How Google Quietly Became An AI Powerhouse
Alphabet's stock has surged 140% in the past year as Google transformed from a perceived AI laggard into a full-stack AI powerhouse. The company leveraged its existing assets—DeepMind, Google Cloud, TPUs, and massive distribution channels—to mount one of the most significant strategic reversals in tech history. Investors now view Alphabet as one of the few companies capable of monetizing AI across every layer of the technology stack.
Summary
The transcript examines how Alphabet went from being seen as a defensive, struggling company in the early AI boom to briefly surpassing Nvidia as the world's most valuable company. Despite having invented the transformer architecture that underlies modern generative AI, acquiring DeepMind in 2014, and being years ahead of competitors, Google was caught flat-footed by ChatGPT's launch in late 2022. This led to a period of intense scrutiny, including the high-profile Bard demo failure, concerns about Gemini's restrictive guardrails, and fears that generative AI would cannibalize Google's core search business.
Google's turnaround began when it stopped competing on OpenAI's terms and instead leveraged its unique full-stack advantages. The company reorganized around AI as its core operating principle, elevating DeepMind from a research lab to a central business engine. Gemini improved significantly, surpassing 750 million monthly active users, and Apple chose Gemini to power a rebooted Siri while also selecting Google as its preferred cloud provider. Google Cloud revenue surged 63% year-over-year in Q1, with 75% of customers using Google's full stack.
A major pillar of the bull case is Google's custom silicon strategy. The company announced two new eighth-generation TPUs in April, one for training and one for inference. Mizuho estimates roughly $61 billion of Google's cloud backlog through 2027 could come from TPU sales. This makes Alphabet a rare AI trade with simultaneous exposure to hardware demand, cloud growth, and model development.
The $200 billion Anthropic cloud contract with Alphabet further strengthened the thesis, as it represents a powerful hedge: even if enterprises choose non-Google AI models, the compute still runs on Google's infrastructure, and Google holds an equity stake in Anthropic, with up to $40 billion committed. However, analysts draw comparisons to Oracle's OpenAI concentration risk and question whether Anthropic could become a similar liability.
Looking ahead, investors are focused on Google I/O for clarity on three fronts: monetizing Gemini, building an agentic commerce strategy, and capturing enterprise AI revenue. Alphabet is projected to spend up to $190 billion in capital expenditures, more than double its 2025 level. Analysts argue that five years from now, 2026 will be seen as still early in the AI cycle, with Google and SpaceX identified as the two most aggressively positioned companies to build intelligence at scale.
Key Insights
- Despite inventing the transformer architecture and acquiring DeepMind years before competitors, Google allowed itself to be disrupted in the very category it pioneered — an analyst argues 'the position that Google got themselves in at the beginning of 2025 should have never happened, because they were years ahead of where their competition was.'
- Google's strategic pivot was not about beating OpenAI at its own game, but rather leveraging its existing ecosystem — an insider notes 'we don't need to be as good as ChatGPT, but we don't want to be ten times worse than ChatGPT,' framing the goal as closing the gap rather than winning outright.
- The Anthropic deal is described as a near-perfect hedge: even if enterprises choose Claude over Gemini, Google still captures revenue through cloud infrastructure, TPU usage, and its equity ownership stake in Anthropic — making the $200 billion commitment a structural win regardless of model preference.
- Mizuho estimates roughly $61 billion of Google's cloud backlog through 2027 could come from TPU sales alone, positioning Alphabet as a rare single-stock exposure to hardware demand, cloud growth, and frontier model development simultaneously.
- An analyst warns that what happened to Oracle — whose stock lost roughly half its value over five months after investors questioned its heavy concentration on OpenAI — now poses a parallel risk for Alphabet if Anthropic transitions from a durable infrastructure flywheel into a customer concentration liability.
Topics
Transcript
[0:00] Alphabet stock is up 140% in the past year and even briefly surpassed Nvidia in early May as the world's most valuable company by market cap, a remarkable feat for a company that was seen as deeply at risk in the early days of the artificial intelligence boom. Now, the rally reflects a major shift in how Wall Street sees Google. A year ago, investors were asking whether generative AI would destroy its search business, and now it's whether Alphabet may be one of the few companies that can actually monetize AI [0:30] across nearly every layer: models, chips, cloud, search, Android and enterprise software. Google, in my opinion, they're probably second best position when it comes to doing…
Full transcript available for MurmurCast members
Sign Up to AccessMore from CNBC
Williams-Sonoma's stock has soared in a sluggish housing market. Here's how it won over Wall Street
Williams-Sonoma has outperformed the S&P 500 and competitors over the past 3 years despite a sluggish housing market by maintaining profitability through premium pricing and reducing reliance on discounts. The company's multi-brand portfolio with omnichannel capabilities gives it a competitive advantage in the fragmented furniture industry.
Why It's So Expensive To Run An Auto Repair Shop
Modern vehicle technology has dramatically increased the costs of running independent auto repair shops, requiring expensive diagnostic computers, software subscriptions, and specialized equipment that dealerships don't face. Motor vehicle repair prices have risen 60% between 2019 and 2026, with independent shops struggling to keep up with fragmented manufacturer systems and rapidly evolving technology.
Take a look inside United's Airbus A321XLR
United Airlines' new Airbus A321XLR features 20 individual Polaris suites with closing doors for premium privacy. The aircraft also includes a snack bar for economy class passengers on long international flights, allowing them to access refreshments at any time.
How Walmart Is Hoping Its New Tech Will Help Shoppers And Employees
Walmart is rolling out digital shelf labels (DSLs) and RFID technology across its stores to improve the shopping and employee experience by enabling faster price updates, easier inventory management, and quicker product location. The technology addresses concerns about dynamic pricing while helping Walmart compete during a period of cautious consumer spending driven by inflation.
How circuit boards could be the next AI chokepoint
Printed circuit boards (PCBs), the foundation for AI chips and electronics, have become a critical supply chain chokepoint as China dominates manufacturing while US demand from AI data centers and defense sectors surges. The US lacks sufficient domestic production capacity to meet growing demand, creating security risks and price pressures.