Thursday, July 16, was supposed to be a quiet day for Alphabet stock. Bloomberg published one story, and by the closing bell, the company had lost roughly $200 billion in market value.
Alphabet shares fell 4.4% that day, closing at $354.46. The trigger was not an earnings miss or a product recall. It was a single report, by Bloomberg reporters Julia Love and Davey Alba, that Gemini 3.5 Pro, Google's flagship AI model, is still failing the company's own internal coding benchmarks months after it was supposed to ship.
Google had said nothing that morning to invite the sell-off. The market did the rest on its own.
A Single Report Moved $200 Billion
Bloomberg's story, citing people familiar with the matter including current and former Google employees, made a simple claim: Gemini 3.5 Pro is not ready, coding is the specific weak point, and Google has known it for weeks. Late in June, according to the report, Google updated the data used to train the model specifically to fix its coding performance. The results disappointed the company's own engineers.
Bloomberg also reported something less technical and more human. "The delay has caused frustration among Google engineers, AI researchers, and managers, who are concerned the company risks losing its edge in the market to rivals Anthropic and OpenAI," the reporters wrote.
That sentence, more than any benchmark score, is likely why the stock moved. Alphabet's AI story has carried a large share of its valuation for two years. A report about internal frustration and a slipping flagship model reads, to investors, like a question about whether that story still holds.
Google's response, given to Bloomberg and repeated to other outlets that asked, did not deny the delay.
"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 on model testing and broader frameworks."
Notice what the statement leaves out: a launch date.
Three Deadlines, One Broken Promise
Gemini 3.5 Pro has been "coming next month" for two months running.
The promise started at Google I/O on May 19, when the company launched the lighter Gemini 3.5 Flash and told developers the more powerful Pro version was close behind. At a media briefing ahead of the keynote, Pichai was specific about timing.
"We are also excited for 3.5 Pro. We are using it internally. It's showing great improvements. We are still testing and refining it, and it will roll out to everyone next month."
"Next month" meant June. June ended with no Gemini 3.5 Pro.
By the timeline several outlets have since reconstructed from Google's public statements and their own reporting, that was only the first slip. A revised target of July 17 began circulating in tech press through early July, though Google never confirmed the date publicly. On July 15, two days before that target, Geeky Gadgets reported, citing the YouTube channel World of AI, that the model had failed to match OpenAI's GPT-5.6 on key benchmarks, citing frequent hallucinations and inconsistent outputs in real-world use. Geeky Gadgets described it as the model's third postponement since the original June target. Bloomberg's story landed the next day, and the market turned a tech-press item into a financial one.
The Rest of the Field Didn't Wait
While Gemini 3.5 Pro sat in testing, three rival labs shipped.
Anthropic returned Claude Fable 5 to general availability on July 1, after a brief export-control review had pulled its two best models offline. OpenAI followed on July 9 with GPT-5.6, which the company said set a new state of the art on the Terminal-Bench 2.1 coding benchmark. A day earlier, SpaceXAI, the merged xAI and SpaceX entity, launched Grok 4.5, which independent benchmarking firm Artificial Analysis ranked fourth on its Intelligence Index. Chinese lab Moonshot AI added a fourth entrant on July 16 with Kimi K3, a 2.8 trillion-parameter open-weight model that early testers said rivaled Fable 5 and GPT-5.6 at a fraction of the cost.
Every one of those models shipped while Gemini 3.5 Pro was still, in Google's own words, being tested with partners.
The timing compounds a separate problem. In June, Google lost four senior Gemini and DeepMind researchers in a single week. Noam Shazeer, a co-author of the Transformer paper and a Gemini co-lead, left for OpenAI. 2024 Nobel laureate John Jumper left for Anthropic days later, followed by Jonas Adler and Alexander Pritzel, also bound for Anthropic. That news alone sent Alphabet shares down more than 5% on June 22, a separate sell-off from Thursday's. A company that is simultaneously behind schedule on its flagship model and losing researchers to the labs it is racing raises an obvious question: is Google still setting the pace here, or just keeping up with it?
Google Says the Numbers Tell a Different Story
Google's position is that a late model is not the same as a failing business, and its own first-quarter numbers back that up.
| Metric | Q1 2026 | Change |
|---|---|---|
| Alphabet revenue | $109.9 billion | Up 22% year over year |
| Google Cloud revenue | $20.0 billion | Up 63% year over year |
| Cloud operating income | $6.6 billion | Roughly tripled year over year |
| Cloud backlog | Over $460 billion | Nearly doubled from prior quarter |
Pichai told investors that Gemini models were processing over 16 billion tokens per minute through direct customer use, up 60% in three months.
"Our AI investments and full stack approach are lighting up every part of the business," Pichai said in the earnings release.
None of that proves Gemini 3.5 Pro will ship soon, or that it will be competitive with GPT-5.6 and Fable 5 when it does. But it is a real argument: customers signed contracts and expanded cloud commitments this spring without waiting for Gemini 3.5 Pro to exist. Google's implicit bet, visible in its own statement to Bloomberg, is that shipping a model late but past its internal quality bar beats shipping one on time that falls short of it. Whether Wall Street agrees depends on how long "late" turns out to be. Alphabet's roughly $190 billion 2026 capital spending plan assumes the answer is not much longer.
The Bottom Line
Strip away the benchmarks and the stock charts, and the story is simple. Google promised its best model for June. It is now the third week of July, three deadlines have passed by outside counts, and the company still will not name a date. Meanwhile, Anthropic, OpenAI, SpaceXAI, and a Chinese lab all shipped competitive models, and four of Google's own researchers left for the two companies it is racing against.
Nothing in Bloomberg's reporting says Gemini 3.5 Pro is a failed product. If anything, it says the opposite: Google is choosing to keep testing rather than ship something that misses its own bar. That is a defensible choice for a company with Alphabet's balance sheet. It is a far harder choice to defend to a market that has priced AI leadership into the stock, and that just erased $200 billion in an afternoon on nothing more than a report that "next month" meant nothing again.
Alphabet reports second-quarter earnings on July 22. The question analysts will ask on that call is not about revenue. It is the same one developers have been asking since May: when.
Sources
- Google Gemini launch delayed as tech falls short of internal goals — Bloomberg, July 16, 2026
- Alphabet shares fall on report its most powerful AI model Gemini 3.5 Pro is delayed — CNBC, July 16, 2026
- Alphabet stock falls on report of Gemini AI model delays — Investing.com, July 16, 2026
- Alphabet Stock Falls 4.4% on Gemini Delay as Buffett Bets on GOOGL — The Cryptonomist, July 17, 2026
- Where is Gemini 3.5 Pro? The AI model announced at Google I/O is still MIA — Mashable via Yahoo Tech, July 17, 2026
- Stopgap Gemini 3.6 Flash May Launch During Gemini 3.5 Pro Delay — Geeky Gadgets, July 15, 2026
- Rebuilt Gemini 3.5 Pro Misses Third Deadline: Google Eyes Stopgap Release — Tech Times, July 16, 2026
- Alphabet's Gemini 3.5 Pro Is Late and the Stock Is Slipping. Is the AI Leader Falling Behind? — The Motley Fool via Yahoo Finance, July 19, 2026
- Google Delays Gemini 3.5 Pro Launch as Coding Performance Doesn't Cut It — Android Headlines, July 18, 2026