Google wants you to know it is still in the AI fight.
The company rolled out three new versions of its Gemini AI model on July 21, a move that tries to fix two problems at once. One of the new models is focused on cybersecurity - a category where rival Anthropic has an early lead with its Mythos model. The other two are designed to be cheaper and faster, giving developers a reason to pick Google over the competition.
The big selling point is money.
That is a direct challenge to a crowded field. Chinese AI companies like Moonshot AI and Alibaba are pushing hard. Moonshot's Kimi K3 recently ran into capacity problems because demand was so high.
Alibaba is teasing a new model called Qwen 3.8 Max that it claims is second only to Anthropic's Fable 5. The message is clear: if you want to win in AI, you either need to be the smartest or the cheapest. Google is betting on cheap.
Google has a history of impressive AI research, but it has also stumbled. The company has faced product delays, and that has given rivals room to move ahead in key areas. Cybersecurity is a perfect example. Anthropic's Mythos model is already out there, and Google has been quiet on that front until now.
Get the market news that matters in a five-minute read with Market Briefs, our free daily newsletter
A Google Cloud spokesperson said the company is "constantly researching and experimenting with new innovations to deliver maximum performance and efficiency for our users and customers." That is corporate-speak for "we are working hard to catch up." The spokesperson also said Google is "co-designing our hardware and software from the ground up" to make sure the systems are integrated and optimized for real workloads.
That hardware comment is a big deal. According to reports, Google is creating a custom chip to improve Gemini's operational efficiency. The goal is a tenfold improvement in efficiency - up to 10 times better than current approaches. That kind of leap would make running AI models much cheaper and could give Google a lasting edge, even if it is behind on timing.
The catch: A custom chip takes years to develop. In the meantime, Google has to fight with the tools it has now. That means cheaper model prices and a broader lineup.
Alphabet, the company behind Google, reports earnings the day after this launch. That timing is not an accident. Investors want to see that Google can keep up in AI without spending money faster than it makes it.
The earnings report will be the real test. Wall Street wants to know if these new models can actually win customers and revenue. In addition, Google is testing Gemini 3.5 Pro with partners and has kicked off its most extensive pretraining effort to date for Gemini 4. That suggests the company is thinking long-term, even as it scrambles to fix short-term gaps.
For investors, the story here is about competition and cost. AI is still a market where every big tech company is spending huge sums. Google is trying to prove it can compete without blowing up its budget.
Cheaper models help, but they also mean thinner profit margins per task. The winner might not be the company with the fanciest model - it might be the one that makes AI cheap enough for everyone to use.
That is the bet Google is placing. If it works, your portfolio - whether you own GOOGL directly or just a broad market fund - benefits from a company that found its footing. If it does not, the competition will keep eating into its lunch. Watch the earnings call for clues on which direction things are heading.
Join Market Briefs, our free daily newsletter, for a quick daily rundown of the markets
