alphabetThe stock is up 140% over the past year, growing faster than the cloud business Amazon’s And Microsoft’s.
But 18 months ago, Google parent looked like it had spent a decade preparing for the artificial intelligence age, only to see OpenAI define the market.
Now, Wall Street is valuing Alphabet as one of the few companies to profit from every layer of the generative AI boom.
Google I/O, which starts Tuesday, has always been a place to show developers where the company is going. This year, the stakes are higher.
Wall Street has already rewarded Alphabet for its AI comeback, but investors want to see whether that confidence is backed by a real product roadmap in key areas like search, cloud, Android, chips and enterprise software.
“Google is probably the best-positioned company to monetize AI at scale because it controls almost every layer of the stack,” said Lou Toney, founding managing partner of Plexo Capital and an early investor in Anthropic. “We’ve never really seen a company that has full vertical integration from top to bottom to be able to support AI.”
Gene Munster, managing partner of Deepwater Asset Management, said the advantage of having multiple layers of control is not just scale, but also speed.
“There’s an advantage to owning the full stack in terms of the speed that you can innovate,” Munster said. “When you’re building on your own custom silicon, for example, it’s a speed advantage. When you have access to power, you can get to data centers more quickly. It’s a speed advantage, which is important.”
Here are seven key areas investors are keeping an eye on at Google I/O:
What’s next for Geminis?
The most closely watched announcement will be whether Google will unveil the next-generation Gemini models.
Reports ahead of I/O have pointed to a possible Gemini 4 debut, although analysts are not entirely confident in this. Citi noted that with Gemini 3.1 Pro released in February, Google has been on a launch cadence of about three to four months, making a Gemini 3.2 or 3.5 update more likely than a full generational leap.
This makes the Gemini 4 question more than a version number. Going one step further would give Google a clean answer to OpenAI and Anthropic. Mizuho wrote that the Gemini 4 announcement would “push Google to the very end of the frontier”, while simply another Generation 3 update would read more like catch-up.
A broader Gemini ecosystem update will also be important.
Mizuho analysts said they will be particularly keeping an eye on progress on Project Astra, Google’s universal AI assistant, as well as deeper Gemini Live capabilities, including screen sharing, video understanding and the use of native tools in Search, Gmail, Calendar and Maps. Updates to Gemma, Google’s open-source model family, and Gemini Robotics are also expected.
Event usage numbers are already stronger than a year ago. Paid Gemini Enterprise monthly active users increased 40% in the first quarter compared to the previous quarter. U.S. monthly active users of the Gemini app increased 127% year-over-year in April, according to Citi data. Token consumption reaches 16 billion per minute according to Google Cloud Next.
AI Agent
If there is one topic running through the I/O session lineup, it is the agent.
Google has sessions on agentic coding workflows, multimodal tools, media generation, robotics, and AI agents. The goal is to position Gemini not just as a chatbot but as an operating layer in Google’s products, capable of understanding context and taking action.
Tony said, “This is what wins the office co-pilot market.” “If the larger market becomes AI agents and orchestrating them — inference infrastructure, multimodal workflows, enterprise search — then that’s where we see a huge opportunity for Google to be able to drive Alphabet’s future growth.”
Agentic coding is part of how Gemini was founded as a response to Anthropic’s Cloud Code and OpenAI’s Codex. That category has become one of the clearest proof points for the commercial value of AI, especially in enterprise software.
agent shopping
Commerce can be a big opportunity. Google already has search, shopping, autofill, and payments; Now he wants Gemini to connect them with an agentic checkout experience.
Google is expanding its Universal Commerce Protocol to include partners metaMicrosoft, Stripe, Klarna And Voice In recent weeks. The I/O is expected to showcase how that infrastructure can enable end-to-end agented checkout, where Gemini not only answers the purchase query but also completes the transaction.
Samir Samat, president of Android ecosystem at Google, asked Gemini to plan a barbecue, create a menu, open Instacart, add ingredients to a Safeway cart, and notify him when the task was completed.
“If you add it up several times a day throughout your week, it adds up to a lot of time,” Samat said. “Those are the features that I think people are much more excited about and are more tangible.”
Tony said Google’s multimodal experience gives it a structural edge as these workflows become more complex.
“Enterprise workflows are increasingly incorporating things like video, voice, images and code,” he said. “Google is uniquely strong in multimodal systems because they have this experience with some of the biggest applications that they handle – YouTube, Android, Maps, Search, DeepMind – and then obviously, TPU.”
For investors, the agentic commerce push has implications beyond Alphabet. Mizuho hints that Google’s more agentic product development could weigh on markets Booking Holdings, Expedia, Doordash, Zillow And instacartGiven that, the anticipation of that change is probably already driving part of the recent weakness in those stocks.
AI mode
The next question is how Google gets paid for this. According to Citi, AI-enabled campaigns now account for more than 30% of search spend. AI Max, which emerged from beta in April and is set to replace dynamic search ads by September, is showing early results, including stronger conversions with its full feature suite.
Citi noted that the AI mode could help Google monetize longer, more complex queries that were historically harder to convert into advertising dollars. But Mizuho points to tradeoffs: AI mode searches are generating far fewer outbound clicks, the firm estimates that 93% end without an outside click, and organic click-through rates on AI observation queries are down 15%.
This makes monetization one of the biggest questions coming into I/O. Munster said he’ll keep an eye on new ad products inside the AI mode, how Google frames agentic commerce, and what it says about more personalized AI experiences.
google cloud
However, for investors, the most consequential I/O announcements may come from cloud and infrastructure.
Cloud has become one of Alphabet’s strongest pillars. It grew 63% year-over-year in the first quarter, outpacing both Azure and AWS. Cloud backlog reached $462 billion, up nearly 90% quarter-on-quarter, half of which is expected to be recognized in the next 24 months. General AI product revenue increased by nearly 800% year-over-year.
CEO Sundar Pichai pointed to several factors behind that growth, including rapidly winning new customers, larger commitments and deeper relationships with existing customers.
On the company’s first-quarter earnings call, he said the number of billion-dollar-plus deals signed through 2025 has surpassed the previous three years combined, while existing clients are coming in more than 30% ahead of their initial commitments.
ai chips
The new wildcard is external TPU sales. Google revealed in the first quarter that it would begin shipping its custom AI chips to external customers in the second half of 2026, with a broader expansion planned for 2027. This is potentially a huge new revenue stream, but investors still don’t know how to model it.
Mizuho wrote that investors will hear detail about whether external TPU sales are recorded as gross sales or royalty revenue, what margins look like and how those deals are accounted for in the backlog.
Tony called TPU one of the least appreciated parts of the Alphabet thesis, arguing that Google’s in-house chips have allowed the company to build a tightly integrated AI infrastructure that supports not only Gemini and the cloud, but also YouTube, Android, and the rest of its ecosystem.
Munster estimates the broader AI chip market is running at about $500 billion annually, meaning even modest share gains could be significant for Alphabet.
anthropic
No relationship will be subject to more scrutiny than I/O Google’s relationship with Anthropic.
Alphabet has a significant ownership stake in AI startups, and the recently reported $200 billion cloud commitment, if accurate, could represent a sizable portion of Google’s contracted future cloud revenues.
Google has also committed up to $40 billion in total investment, creating a loop in which capital flows into Anthropic and back into Google through cloud and TPU spending.
This dynamic raises a concentration question that investors have already seen elsewhere in the cloud. OracleIts stock was largely tied to OpenAI after it reported a massive backlog surge, then sold off as investors became more concerned about customer concentration. Microsoft faces a similar debate over its OpenAI relationship.
Tony said the Google-Anthrop relationship looks more like a defense than a weakness.
Even if enterprises choose the cloud instead of Gemini, Google could still benefit from demand for the infrastructure behind that use, he said.
“If enterprises prefer the cloud, Google still wins in infrastructure because all that activity has to live somewhere,” he said. “Google still wins because of its TPU.”
Watch: Inside Alphabet’s capital expenditure payout: What investors need to know
