Asger Risom
For three years after ChatGPT’s debut, Google was cast as the giant that had missed the moment — but not anymore.
For three years after ChatGPT’s debut, Google was cast as the giant that had missed the moment. Analysts, rival engineers and even former executives questioned whether the company could keep pace with OpenAI. Now, with a string of model releases, swelling chip demand and a surge in investor confidence, the narrative has shifted.
The giant reawakens
Google’s latest model, Gemini 3, won immediate praise for its reasoning, coding performance and ability to handle tasks that have tripped up rival chatbots. The launch coincided with new partnerships — including a major TPU deal with Anthropic — and rising demand for Google’s custom chips, one of the few alternatives to Nvidia’s dominant hardware.
Alphabet has added nearly $1 trillion in market value since mid-October, buoyed by Warren Buffett’s investment and strong cloud growth linked to AI. Reports that Meta is exploring Google’s processors for future data centers pushed Alphabet’s shares toward a record $4 trillion valuation.
“It’s a sleeping giant that is now fully awake,” Neil Shah of Counterpoint Research told Bloomberg.
A full-stack strategy
Executives have argued that Google’s strength lies in its ability to build every layer of AI computing: applications like Gemini, the models beneath them, the cloud infrastructure supporting them and the in-house TPUs that power training. “We’ve taken a full, deep, full-stack approach to AI,” Sundar Pichai said last quarter.
That integrated system gives Google control and reduces reliance on external suppliers. It also gives the company a vast dataset drawn from Search, Android and YouTube — resources OpenAI and other rivals cannot match.
The surge in TPU interest marks a shift. For years, Google was effectively the only buyer of its processors. Anthropic’s agreement to use up to 1 million TPUs, and Meta’s reported interest, indicate broader adoption.
Reorganizing for the fight
Google’s resurgence is tied to its 2023 consolidation of AI research under DeepMind cofounder Demis Hassabis. Despite missteps, including a flawed image-generation rollout, the reorganization sharpened the company’s focus on foundational models capable of competing directly with OpenAI and Microsoft.
Gemini 3 Pro now sits atop AI leaderboards such as LMArena, with early testers like Andrej Karpathy calling it “clearly a tier 1 LLM.” Google says 650 million people use the Gemini app, though OpenAI’s ChatGPT still holds a larger share of consumer usage.
Market pressure and new risks
Google’s cloud division posted a 34% annual revenue increase, though it remains behind AWS and Microsoft Azure. Analysts say enterprise AI adoption still leans toward competitors, but Google’s chip ecosystem is changing perceptions. Some warn, however, that TPUs lock customers into Google’s cloud, limiting flexibility compared to Nvidia GPUs.
Still, confidence is rising. “Google is back in the game with Gemini 3,” Forrester analyst Thomas Husson said, adding that reports of the company’s decline “have been widely exaggerated.”
Bloomberg, Fortune
