After wrapping up its longest monthly losing streak on Wall Street in more than a decade, Google has started September by giving investors several reasons for optimism.
The company on Wednesday launched Gemini 3.8 Flash, its third Flash model in six weeks, alongside a new cybersecurity model aimed at trusted government and enterprise customers. Berkshire Hathaway CEO Greg Abel also offered a public vote of confidence for Alphabet's AI position.
Meanwhile, a federal judge on Wednesday rejected the Justice Department's push to force Google to sell its ad exchange, another victory for the company against the government's antitrust efforts.
Taken together, the developments offer a more constructive setup for Alphabet after a difficult summer, when Google lost some of its momentum in the AI model race, saw high-profile talent departures, and underwent a major restructuring inside DeepMind.
The stock hasn't yet shown much of a rebound. It rose 0.6% on Wednesday after dropping more than 1% Tuesday to start the month, so it's still down slightly as September gets going.
But Google is enthusiastic about Gemini 3.8 Flash, which is focused heavily on coding and agentic tasks — two of the areas where AI companies are increasingly trying to turn technological advances into enterprise revenue. Google calls it Gemini's best reasoning and coding model yet, with significant improvements over 3.7 Flash in software engineering and multi-step tasks.
Tulsee Doshi, senior director of product management at Google DeepMind, told CNBC the recent Flash models have "really surprised us in positive ways in their performance," adding that they "give us opportunities to lean into them."
Smaller Flash models are cheaper to run and faster to iterate than the company's largest frontier systems, while increasingly approaching them on some tasks. But it's not enough to get the company close to Anthropic or OpenAI, said Gil Luria, an analyst at D.A. Davidson, in an interview.
"From a product perspective this model seems to keep Google in the race, but probably won't change the fact they are a distant third in the enterprise market," said Luria, who recommends holding the stock.
Competing on price
Gemini 3.8 Flash costs 75 cents per million input tokens and $3.75 per million output tokens — the same introductory price as its last Flash model, even with the new one offering improvements in coding, agentic tasks and reasoning.
Gemini Enterprise is adding pay-as-you-go pricing, token discounts of up to 20%, monthly caps on agent spending and a zero-dollar base subscription option. According to materials provided to CNBC, Google has also directly targeted Microsoft and Anthropic, claiming their recurring seat fees and separate product licenses make their offerings more expensive and less flexible.
Google also has the benefit of scale. Nearly three-quarters of Google Cloud customers are already using its AI products, and Google Cloud CEO Thomas Kurian told CNBC that those customers are spending about 50% more than their original commitments.
DeepMind's Demis Hassabis has similarly described a future that plays to that strategy. Hassabis told the G20 Innovation meeting on Wednesday that Gemini could increasingly serve as a general-purpose layer coordinating cheaper, specialized models, and agents. In that world, breadth could matter as much as having the best model.
It was the first time to hear Hassabis speak publicly since DeepMind's reorganization last month. As part of that announcement, Hassabis moved from CEO of DeepMind to chairman of the unit.
Google is also touting costs in cybersecurity. The company says Gemini 3.8 Flash Cyber can detect and patch software vulnerabilities at frontier-level performance while running substantially faster and more affordably than larger models.
"We're really excited about being able to provide an offering to defenders that is a fraction of the cost, much faster, while still showcasing that frontier-level performance," Doshi said.
Because those capabilities could also be misused, Google is initially limiting access to a small group of trusted government and enterprise defenders through its new Fairwind Program.
There's a tradeoff for investors. Alphabet is spending enormous sums on AI infrastructure, so it's counting on growth and market share gains to deliver returns over the longer term.
Berkshire's Abel told CNBC's Becky Quick on Wednesday that it sees Alphabet as a winner in AI, a view partly based on how its portfolio companies use its technology.
"We have a lot of visibility from within our companies as to how we're using AI, what type of benefits it's delivering, so that brought incremental interest, and then we saw Google as a significant player," Abel said.
While Google's position in AI is viewed as tenuous by some market experts, the ad business keeps on humming along, growing 14% in the latest quarter.
That cash engine received another boost on Wednesday.
A federal judge overseeing the Justice Department's ad-tech antitrust case ruled that Google will not have to sell its AdX exchange, opting for behavioral remedies rather than the structural breakup regulators had sought. The decision follows a separate antitrust ruling last year, when a judge rejected calls to force Google to divest Chrome.
Antitrust attorney Wyatt Fore, a partner at Schinder Cantor Lerner, told CNBC it's a "big deal" that courts in both cases opted against breaking up the company, adding that Google is heading into the AI race with "no hands tied behind its back."
WATCH: Google's Pixel 11 puts Gemini Intelligence at the center of the smartphone