Analysis
Alphabet shares fell 4.4% on July 16, wiping out roughly $200 billion in market value in a single session, after Bloomberg reported that Gemini 3.5 Pro -- the model Google unveiled in May as its most powerful ever -- is running months behind its promised June rollout. The company had billed the launch as an aggressive follow-up to Gemini 3, promising broad public availability within a month of the May unveiling. That deadline came and went without comment.
The specific cause, according to Bloomberg's sourcing, is that Gemini 3.5 Pro's coding performance fell short of Google's internal targets, and the company chose to keep iterating rather than ship a model that would immediately be compared unfavorably to OpenAI's GPT-5.6 Sol and newer Meta systems, both of which have posted stronger code-generation benchmarks in recent months. Coding capability has become the single most closely watched proxy for frontier-model quality this year, in large part because it's the use case enterprises are willing to pay the most for.
“The timing made the market reaction sharper than a routine delay would normally warrant.”
The timing made the market reaction sharper than a routine delay would normally warrant. Google's stumble landed in the same 24-48 hour window as Moonshot AI's Kimi K3 reveal, which triggered a broader selloff in AI-adjacent equities on fears that the sector's spending has outrun its output. A delayed flagship model from the company with the largest AI infrastructure footprint reinforced exactly the narrative investors were already nervous about: that execution risk across the entire AI stack is higher than the last two years of triple-digit-multiple valuations assumed.
For enterprise buyers and the VCs backing companies built on top of Gemini's API, a multi-month delay on the flagship model means more time for OpenAI and Anthropic to consolidate developer mindshare -- and more pressure on Google to eventually ship something that clears a higher bar than it would have needed to in June. What to watch next: whether Google gives any public timeline update before its next earnings call, and whether the delay shows up in developer-platform market-share numbers before the model actually ships.
