Google CEO Sundar Pichai is likely to face sharp questions during upcoming Q2 earnings about a simple problem: Google is still delaying its frontier model, Gemini 3.5 Pro, even as rivals move fast. The timing is where the pressure is coming from. This week, Google rolled out three faster, more cost-effective models, but Gemini 3.5 Pro remains “next,” and the interim looks like an opening competitors are not wasting.

The jabs are already out in the open. Meta’s chief AI officer, Alexandr Wang, posted “gemini who?” on X in response to a leaderboard that ranked Meta’s Spark model above one released by Google this week. A member of technical staff at OpenAI, Thibault Sottiaux, also took an apparent swipe, replying to a Google-related update with “Hope it finishes one day too!” Logan Kilpatrick, who announced on X that pre-training on Gemini 4, the next major milestone model, had begun. Google declined to comment.

This is a classic AI-race dynamic, but the new twist is how publicly and quickly it plays out. In theory, Google’s strategy is coherent: it has been pushing for “faster, more cost-effective models” as token costs become a real business constraint for anyone building or deploying AI. The source notes that analysts and users have praised this efficiency angle, including a founder who said Gemini 3.5 Flash is his “daily driver” for agentic document extraction. That matters because efficiency is not just a tech preference. It directly affects whether models are economically usable at scale for workflows like extraction.

But “economically usable” and “frontier leader” are two different categories, and the market has started treating them that way. One analyst, Josh Beck at Raymond James, said the delay has “shifted perception from leading edge to trailing edge.” His point is not that Google stopped moving. It is that, in a fast cycle where labs ship top-tier models in “recent weeks,” a delay can change what people think is happening. Perception drives adoption. Adoption drives compute spending. Compute spending is the oxygen of the entire ecosystem.

The pressure becomes more immediate because Google is not competing only with one lab. The article frames the issue against OpenAI and Anthropic rolling out new top-tier models in recent weeks. When those releases stack up while Google’s next frontier offering keeps slipping, the narrative gets easier for competitors to write. And competitors are writing it on X, not in private board decks. That is a problem for a company that still has to answer questions in investor settings, where “timeline uncertainty” can be as damaging as weak performance.

Analysts also expect the topic to land during Google’s Q2 earnings on Wednesday evening. They are likely to raise 3.5 Pro and its release timeline, because Google is basically asking the market to accept a trade-off. Either the frontier power is coming later, or the company is betting the market will value speed and cost over peak capability right now. Some users are buying the efficiency story. Still, the trade-off can look like a stall if the promised “next” model is not actually shipping.

The source also adds another layer: Google is hyping Gemini 4 while 3.5 Pro is delayed, and that has some analysts reading it as a signal. Moor Insights & Strategy analyst Anshel Sag wrote on X that Google hyping 4 “before even delivering 3.5 Pro is a lil weird.” Sag told Business Insider he felt it was an “admission they already have something better,” though he also cautioned that the “feverish pace” of AI may not guarantee meaningful improvements right away. His conclusion is more nuanced than the online jabs: it’s “way too early to count anyone out.” That nuance is important because it reflects how boards and investors may interpret this. Delays can be technical, not strategic. But they still need explaining.

For decision-makers, the second-order problem is not only Google’s product cadence. It is how quickly the AI category now turns delivery timing into a competitive weapon. Meta’s “gemini who?” isn’t about one model leaderboard entry. It is about positioning. OpenAI’s technical staff does not have to persuade builders on the spot. They just have to keep Google in the penalty box until the next release arrives. Meanwhile, Google’s internal bet is that faster, more cost-effective models can “win” in the real world while the frontier capability catches up behind them.

So the stakes for Sundar Pichai, and for any board watching this sector closely, are simple: when the next frontier model is delayed, every earnings call becomes a referendum on credibility. Google can still be right about efficiency. It can still be right about long-term roadmap execution. But in the short term, the market will ask whether the gap between “leading edge” and “trailing edge” is closing fast enough to justify the wait. And until Gemini 3.5 Pro is in the hands of users, rivals will keep filling the silence with punchlines.