Amazon Web Services hit a big milestone by reaching $123 billion in yearly revenue. The cloud computing giant made $30.9 billion in the second quarter of 2025, which is 17.5 percent more than the same time last year.

But there's a catch that has investors worried. While AWS still leads the pack in total revenue, it's growing slower than its main rivals. Google Cloud grew 32 percent compared to last year, and Microsoft grew 26 percent. AWS only grew 17.5 percent.

The company is spending massive amounts of money to stay competitive. Amazon spent $31.4 billion on new equipment and buildings this quarter alone. For the whole year, they expect to spend more than $118 billion. That's way more than Google's $22.4 billion or Microsoft's $24.2 billion per quarter.

Most of this money goes toward building new data centers and buying computer chips for artificial intelligence work. Despite all this spending, AWS still can't keep up with demand. Jassy said power supply is their biggest problem, followed by chip shortages and server parts.

The heavy spending is hurting AWS's profits. Their profit margins dropped from 39.5 percent to 32.9 percent between the first and second quarters.

When asked about falling behind in AI compared to competitors, Jassy said it's still very early in the AI race. He pointed to AWS's partnership with AI company Anthropic and their own custom computer chips that offer better value for money.

Microsoft just shared some big numbers that show how much money their cloud business is making. Their Azure cloud service alone brought in $75 billion this year. That's a huge amount of money from just one part of their business.

The company built a lot of new data centers too. They added more than 2 gigawatts of data center power in the past year. To put that in simple terms, that's enough electricity to power about 1.5 million homes.

Microsoft now runs more than 400 data centers around the world in 70 different areas. All these new data centers are built to handle AI work and use special liquid cooling systems.

Even with all this building, Microsoft still doesn't have enough space for all the customers who want to use their services. They're what you call "capacity-constrained," which means demand is higher than what they can supply.

The company spent $24.2 billion in the last three months of their fiscal year just on building new facilities and buying equipment. That's 27 percent more than they spent in the same period last year. Next quarter, they plan to spend even more - around $30 billion.

Their CFO Amy Hood explained that they have $368 billion worth of contracts already signed with customers. This means they know they'll have plenty of business to fill all the new data centers they're building.

For the whole company, Microsoft made $76.4 billion in revenue this quarter, which is 18 percent more than last year. Their cloud division brought in $29.9 billion of that total.

Ron Gula from Tenable used to explain there are two types of free in open source: free as in freedom and free as in beer. Most users only care about the beer part - they want to use the software without paying.

Major cloud projects like Prometheus, OpenTelemetry, and Jenkins power critical business systems worldwide. Yet many of these projects struggle with money problems. Running big projects can cost hundreds of thousands of dollars each year just for hosting.

The Linux Foundation and CNCF try to help by supporting over 200 open source projects. But even they face limits on what they can fund.

Developers get tired and quit. Projects lose support when companies can't make money. Security updates stop coming. Documentation becomes outdated. When maintainers disappear, critical software can break.

Some companies try to solve this through paid enterprise versions or support services. But most projects still struggle to find ways to make money while keeping their software free.

The solution requires a culture change. Companies need to support the projects they use.

Google just rolled out version 0.3 of its Agent2Agent protocol, making it easier for companies to use AI agents that can talk to each other.

The biggest change is adding gRPC support. Think of gRPC as a super-fast way for computer programs to chat with each other, even when they're running on different machines or written in different coding languages. This means AI agents can now work together much faster and smoother.

Google also added better security features. Companies can now make sure their AI agents are who they say they are by using digital signatures. This is huge for big companies who won't use AI agents unless they can prove they're safe and trustworthy.

The company made it work better with their Agent Development Kit too. Now when developers build AI agents using Google's tools, those agents automatically get the ability to communicate using the new protocol. It's like getting a built-in translator for free.

There's competition though. MCP is also trying to become the standard way for AI agents to communicate. Experts say MCP is easier to use and works with more different systems. But Google's version is more secure and better for big companies.

Google has about 150 partners helping them push their protocol, while MCP seems more popular with individual developers. MCP works great when you need one agent to use many different tools. Google's version shines when you need many agents working together on complex tasks.

Rick Ochs teaches us that Finops and Cost Optimization for AWS is not just technical but also cultural. To succeed, it is essential to balance performance, costs, and collaboration, driving a mindset shift within organizations.

Michael Wittig noticed something strange happening with his AWS bill. Each month for over a year, he got budget alerts for his AWS account that hosts their monitoring solution called marbot.

At first, the numbers seemed to make sense. His service usage grew by 33% and his AWS bill went up by 20%. But when he looked closer, he found the real problem. AWS Config costs had jumped by 1400% from $6 to $90 per month.

Here's what was happening. Every time a Fargate task started, it created a network interface that triggered 4 different configuration changes to be recorded. Those 4 changes cost $0.012, while running the actual Fargate task for a minute only cost $0.0008.

Michael couldn't just turn off AWS Config because his security audit needed it. But he found a simple fix. Instead of recording changes continuously, he switched to recording them daily.

The change was simple. He updated his CloudFormation template to set the recording frequency to daily instead of continuous. This one small change saved him $90 per month or $1080 per year.

A hacker successfully inserted harmful code into Amazon's Q coding assistant tool and it was sent out to nearly a million users through an official update.

The attacker managed to get into Amazon Q's extension for Visual Studio Code, which has been downloaded over 950,000 times. They used a fake GitHub account to submit changes in late June and somehow got admin access to the system.

On July 13, the hacker added bad code that told the AI tool to act like a system cleaner. The code was designed to delete user files and cloud data. Amazon released this broken version on July 17 without knowing it had been changed.

The hacker told reporters they could have done much worse damage but chose to keep it limited as a way to protest Amazon's weak AI security measures.

Amazon quickly fixed the problem and said no customer data was actually harmed. They released a clean version and told users they could update to be safe.

Security experts say this shows how dangerous it can be when AI coding tools get hacked. When these systems are broken into, bad actors can put harmful code into software that millions of people use. Users mig

Join FinOps professionals at the FinOps Weekly Summit 2025 and discover how to:

Transform from reactive fire-fighting to strategic leadership — Learn the proven frameworks that top practitioners use to turn cloud cost management into a competitive advantage

Master AI-powered cost optimization — Get exclusive access to the latest automation tools and techniques that can reduce your cloud spend by up to 40% while accelerating innovation

Build bulletproof unit economics — Walk away with actionable playbooks for calculating true cloud ROI and proving business value to executives who control your budget

"But I don't have time for another webinar..."

This isn't another generic webinar. This is the only major FinOps event left in 2025, featuring battle-tested strategies from companies managing billions in cloud spend.

October 23rd, 2025 | 4:00 PM - 8:00 PM CEST

Limited seats available