Alphabet Stock Drops Nearly 5% After Report Says Google's Top AI Model Is Months Late

Alphabet had a rough Thursday. Shares sank nearly 5% intraday and closed down roughly 4.5% to 4.75%, landing around $353 to $354 a share, according to Benzinga. The trigger was a Bloomberg News report saying Google's flagship AI model, Gemini 3.5 Pro, is months behind schedule.

CEO Sundar Pichai announced Gemini 3.5 Pro back in May at Google's I/O developer conference, saying it was already being used internally and would get a wider rollout the following month. That didn't happen. Bloomberg, citing 10 current and former employees, reported the model's coding capabilities came in short of Google's own internal targets, at the exact moment rivals are eating Google's lunch in that category.

Coding has become the hot battleground in AI right now. OpenAI released GPT-5.6 Sol last week, with CEO Sam Altman touting it as 54% more token-efficient on agentic coding tasks. Meta rolled out Muse Spark 1.1, which AI chief Alexandr Wang called the company's "strongest model for agentic and coding work yet." Anthropic's models are also in the mix. Chinese labs like Z.ai are pushing free, open-weight coding models into the market too. Google, by Bloomberg's account, is watching that gap widen instead of close.

Why It's Reportedly Stuck

Benzinga's reporting on the Bloomberg story points to something more structural than a simple bug fix: organizational sprawl. Multiple Google teams, DeepMind, Cloud, Android, and Search, are all reportedly building overlapping AI coding tools at once, with shifting priorities slowing everything down. That's a classic big-company problem. When five departments are all trying to own the same product, you get meetings, not shipping.

Google tried to fix the coding shortfall in June by updating the training data feeding Gemini, according to Bloomberg's sourcing repeated across Stocktwits and the Benzinga writeup. That update reportedly didn't move the needle enough. Employees and researchers inside Google are said to be frustrated, worried the company is ceding ground in one of the most commercially important parts of the AI race.

Google's Response

A Google spokesperson gave nearly identical statements to CNBC, Reuters, and Bloomberg: "We're shipping quickly across a wide range of models while keeping them highly cost-effective for customers. We're currently testing 3.5 Pro, an upgraded Flash model, and other models with partners, and we're productively engaged with the U.S. government."

Google isn't saying the model doesn't exist or that it's been scrapped. It's saying it's in active testing with partners and the company is deliberately prioritizing cost efficiency over a rushed release. A company betting its AI franchise on getting coding right has an obvious incentive to delay rather than ship something half-baked, especially after watching competitors take heat for premature launches.

This isn't the only AI safety-related delay happening industrywide. Reuters reported, via Euronext's wire feed, that OpenAI's GPT-5.6 launch itself got held up earlier by U.S. government national security concerns over misuse of powerful AI models. Anthropic had its own advanced models, Mythos 5 and Fable 5, disabled entirely for users after a June 12 export control order, before the government lifted those restrictions in late June once Anthropic added safeguards. Delays tied to government scrutiny aren't unique to Google right now.

What the Market Did With It

Stocktwits reported something interesting that CNBC and Benzinga didn't dwell on: retail traders on the platform largely shrugged off the selloff, with some calling it a buying opportunity. Shares actually trimmed some losses after hours, up slightly in extended trading per Stocktwits data, after the roughly 4.5% regular-session drop. A near-5% one-day stock decline is real money wiped off Alphabet's market value for a few hours, but it is not evidence that Wall Street or retail believes Google's AI business is collapsing.

The unresolved question is whether this is a temporary stumble or the start of a pattern. Google has the cash, the compute, and the research talent to eventually ship a competitive coding model. What it apparently doesn't have right now is a lean enough org chart to move as fast as OpenAI or Anthropic. Bloomberg's sourcing suggests Google engineers themselves are the ones sounding the alarm on that point. No specific new release date for Gemini 3.5 Pro has been announced.