In the early days of September 2026, Google emerged from a bruising summer with two developments that, while modest in isolation, carry the weight of institutional survival: a new AI model built for the domains where technology meets commerce, and a court ruling that left its most profitable business intact. The company is not claiming to lead the AI race, but rather to remain in it — competing through price, scale, and the quiet leverage of customers already inside its ecosystem. Whether these are the first steps of a genuine recovery or a temporary reprieve from deeper structural challenges
Google Regains Momentum with New AI Models, Antitrust Win
keeping Google in the race, but probably won't change the fact they are a distant third
Mark
So Google had a rough summer, and now they're announcing new models and winning a court case. Are these actually significant developments, or is this just noise?
Mimi
Both, maybe. The antitrust win is genuinely important—it means Google keeps its ad business intact and doesn't have to break itself apart. That's real. The new AI model is real too, but it's more of a continuation than a breakthrough. They're saying it's better at coding and reasoning, but they're not claiming it beats OpenAI or Anthropic.
Luke
Right, and we should be clear about what the analyst said. Luria called Google a "distant third" in enterprise. That's not a small gap. The pricing is aggressive, but cheaper doesn't automatically win if the product isn't perceived as best-in-class.
Mimi
True. But Google's angle is different. They're not trying to be best. They're trying to be the layer that coordinates everything—the operating system for AI, in a way. And they have three-quarters of Google Cloud customers already using their AI products.
Mark
So they're leveraging existing relationships rather than winning new customers on merit?
Mimi
Not instead of—in addition to. But yes, that's a real advantage. And the ad business is still growing 14 percent, which funds all of this.
Luke
That's the thing though. We don't know if that ad growth continues if the AI bet doesn't pay off. The company is spending enormous sums on infrastructure. At some point, investors will want to see returns, not just customer spending increases.
Mark
What about the cybersecurity model? That seemed like a specific use case.
Mimi
It is. They're saying it performs at frontier level but costs a fraction as much and runs faster. But they're restricting access to trusted government and enterprise customers through this Fairwind program, so it's not a broad market play yet.
Luke
And we should note—that restriction is partly because the capabilities could be misused. So there's a real limit to how much they can scale that product.
Mark
So the headline is recovery, but the reality is more complicated?
Mimi
The headline is that Google stopped the bleeding and showed it's still in the game. The reality is they're in third place and betting on a strategy that hasn't been proven yet. The antitrust win is real relief, though.
Luke
And the stock barely moved. That tells you something about how much confidence the market actually has in these announcements.
The Pulse
- After its worst Wall Street losing streak in over a decade, Google needed proof it could still compete — and September delivered two pieces of evidence, however incomplete.
- The release of Gemini 3.8 Flash, its third Flash model in six weeks, signals a company accelerating not because it is winning, but because it cannot afford to fall further behind OpenAI and Anthropic.
- A federal judge's refusal to force the sale of Google's ad exchange removed a structural threat, preserving the cash engine that funds the company's enormous AI infrastructure bets.
- Google is not competing on having the best model — it is competing on price, flexibility, and the fact that three-quarters of its cloud customers are already using its AI products.
- Analysts remain unconvinced: one prominent voice called the new model enough to keep Google 'in the race' while confirming it remains a distant third in enterprise AI, and the stock's 0.6 percent gain barely registered.
In the early days of September 2026, Google emerged from a bruising summer with two developments that, while modest in isolation, carry the weight of institutional survival: a new AI model built for the domains where technology meets commerce, and a court ruling that left its most profitable business intact. The company is not claiming to lead the AI race, but rather to remain in it — competing through price, scale, and the quiet leverage of customers already inside its ecosystem. Whether these are the first steps of a genuine recovery or a temporary reprieve from deeper structural challenges is a question the market has not yet answered.
Google entered September carrying the weight of a difficult summer — a prolonged stock decline, departures of prominent AI talent, and a visible lag behind OpenAI and Anthropic in the enterprise market. On a single Wednesday, the company released Gemini 3.8 Flash, its third Flash model iteration in six weeks, alongside a specialized cybersecurity variant, and received a favorable federal court ruling that spared its ad exchange from a forced sale.
The new model targets coding and agentic tasks — the specific capabilities AI companies believe will convert technological promise into paying customers. Google claims it represents the company's strongest reasoning and coding model to date, yet the strategy is not to win on raw performance. Instead, Google is competing on price and flexibility: Gemini 3.8 Flash carries the same introductory cost as its predecessor despite performance gains, and new enterprise pricing structures include pay-as-you-go options, token discounts, and a zero-dollar base tier. In marketing materials, Google directly compared its costs to those of Microsoft and Anthropic, framing rivals' seat fees and separate licenses as burdensome by contrast.
The company's deeper advantage may be structural. Nearly three-quarters of Google Cloud customers already use its AI products, and those customers are spending roughly 50 percent more than originally committed. Demis Hassabis, recently elevated to chairman of DeepMind, described a vision of Gemini as a coordination layer — orchestrating cheaper, specialized models rather than simply being the most powerful one. The cybersecurity model, restricted initially to vetted government and enterprise defenders, exemplifies this approach: frontier-level detection at a fraction of the cost and speed of larger models.
The antitrust ruling added a separate dimension of relief. By imposing behavioral restrictions rather than a forced divestiture of its ad exchange, the court left Google's most profitable business — which grew 14 percent last quarter — intact to fund its AI ambitions. An antitrust attorney called the outcome 'a big deal,' observing that Google enters the AI race with no structural constraints on its resources. Even Berkshire Hathaway's Greg Abel offered a public endorsement, citing visibility from portfolio companies into how Google's technology is being deployed.
Yet the market's response was telling in its restraint: shares rose just 0.6 percent. Analysts acknowledged that the new model keeps Google competitive without changing its standing as a distant third in enterprise AI. The company is spending heavily on infrastructure, betting that scale, pricing discipline, and existing customer relationships will eventually translate into the kind of dominance it once held in search. Whether that bet pays off remains the defining question of Google's next chapter.
Google arrived at September with two pieces of good news it desperately needed. On Wednesday, the company released Gemini 3.8 Flash, the third iteration of its Flash model line in six weeks, along with a specialized cybersecurity version designed for government and enterprise use. That same day, a federal judge rejected the Justice Department's demand that Google divest its ad exchange, AdX, choosing instead to impose behavioral restrictions on how the company operates the platform. These developments arrived after what Google's leadership would prefer to forget: the company's longest monthly losing streak on Wall Street in more than a decade, a summer marked by departures of high-profile AI talent and a visible stumble in its race against OpenAI and Anthropic.
The stock market's response was muted. Shares rose 0.6 percent on Wednesday, a modest gain that barely offset Tuesday's decline. Yet the company is framing these moments as a turning point. Tulsee Doshi, senior director of product management at Google DeepMind, told CNBC that the recent Flash models have "really surprised us in positive ways in their performance," and that they "give us opportunities to lean into them." The new model emphasizes coding and agentic tasks—the specific domains where AI companies believe they can convert technological capability into paying customers. Google claims Gemini 3.8 Flash represents the company's best reasoning and coding model to date, with substantial improvements over its predecessor in software engineering and multi-step problem-solving.
Google's strategy, however, is not to win on raw capability. Instead, the company is betting on price and scale. Gemini 3.8 Flash costs 75 cents per million input tokens and $3.75 per million output tokens—the same introductory price as the previous Flash model, despite the performance gains. The company has introduced new pricing options for Gemini Enterprise, including pay-as-you-go structures, token discounts reaching 20 percent, monthly spending caps for AI agents, and a zero-dollar base subscription tier. In marketing materials provided to CNBC, Google directly compared its offerings to those of Microsoft and Anthropic, arguing that their recurring seat fees and separate product licenses make them more expensive and less flexible. Gil Luria, an analyst at D.A. Davidson, acknowledged the strategy but remained skeptical of its ultimate impact. "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," Luria said, recommending investors hold the stock.
Google's structural advantage lies in its existing customer base. Nearly three-quarters of Google Cloud customers are already using its AI products, and Google Cloud CEO Thomas Kurian reported that these customers are spending roughly 50 percent more than their original commitments. Demis Hassabis, who recently transitioned from CEO of DeepMind to chairman of the unit as part of a broader reorganization, articulated a vision that plays to this strength. Speaking at a G20 Innovation meeting on Wednesday, Hassabis described a future in which Gemini functions as a general-purpose coordination layer, orchestrating cheaper, specialized models and agents. In such an arrangement, breadth of offering could prove as valuable as having the single best model.
The cybersecurity model, Gemini 3.8 Flash Cyber, exemplifies this approach. Google claims it 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 these capabilities could be misused, Google is initially restricting access through a new program called Fairwind, limiting it to a vetted group of government and enterprise defenders.
The antitrust ruling provided a separate but significant boost. The federal judge overseeing the Justice Department's ad-tech case determined that Google would not be forced to sell AdX, opting instead for behavioral remedies. This decision echoes a prior ruling in which a judge rejected calls to force Google to divest Chrome. Wyatt Fore, an antitrust attorney at Schinder Cantor Lerner, called the outcome "a big deal," noting that Google is entering the AI race with "no hands tied behind its back." The ad business itself remains robust, growing 14 percent in the latest quarter, providing the cash flow that funds Google's enormous infrastructure spending on AI.
Berkshire Hathaway CEO Greg Abel offered a public endorsement of Google's AI position on Wednesday, telling CNBC's Becky Quick that Berkshire sees Alphabet as a winner in the space. Abel pointed to visibility from within Berkshire's portfolio companies regarding how they deploy Google's technology and the benefits it delivers. Yet the fundamental tension remains unresolved: Alphabet is spending vast sums on AI infrastructure, betting that growth and market share gains will eventually justify the investment. Whether the company can convert its scale, pricing discipline, and existing customer relationships into the kind of enterprise dominance it once held in search remains the open question as the AI market continues to take shape.
Notable Quotes
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— Gil Luria, analyst at D.A. Davidson
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— Tulsee Doshi, senior director of product management at Google DeepMind
Want the full story?
Read the original at CNBC ↗
Technology