From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.

IT Brew

Tom McKay

itbrew.com

DNA strand next to a copy-paste select window

DNA storage could be the solution to future enterprise archival needs

DNA is dense, long-lasting, and stable—but scientists are still trying to figure out how to read and write it quickly.

Nvidia Blackwell chips are next hot commodity for megascalers

Megascale customers line up for Nvidia’s Blackwell AI chips.

Billy Hurley

Green transparent cloud silhouette with binary code displayed behind

Asked and answered: Should in-cloud data be backed up on-prem?

A round-up of industry pros offer some cloud recommendations.

Cloud

Brianna Monsanto

Amazon Web Services (AWS) wants to make building and deploying advanced AI agents feel like a walk in the park.

 annual Summit in New York City in mid-July, the cloud giant unveiled Amazon Bedrock AgentCore, a suite of services that aim to take on the “undifferentiated heavy lifting” of building 

 at scale. Announced AgentCore tools included:

A serverless runtime environment that is framework and protocol agnostic.

 A designated identity and credential management service for AI agents.

 A service that allows existing APIs and Lambda functions to be made into “agent-ready” tools “with just a few lines of code.”

A tool that enables developers to “trace, debug, and monitor” their AI agent’s performance.

A service with both short-term and long-term memory capabilities that allow developers control over the context their agent has.

A service that allows AI agents to write and execute code in a sandbox environment.

 A “secure, isolated” cloud-based browser environment that allows agents to interact with websites.

The set of services, currently available in preview, can either be used in its entirety or on an individual basis. Swami Sivasubramanian, AWS’s VP for agentic AI, said in his keynote that AgentCore was a “step change” for securely moving 

“It provides tools and capabilities to make agents more effective and more capable, purpose-built infra to securely scale agents and provide the controls to enable you to build trustworthy agents,” Sivasubramanian said during the July 16 keynote.

In addition to AgentCore, AWS announced that AI agents and tools from its partners will now be available on the AWS marketplace. The move will allow customers to purchase agentic solutions with “streamlined procurement and multiple deployment options.”

“With this launch, you have access to a centralized catalog of AI agents, tools, solutions, and services from popular vendors,” Sivasubramanian said.

 The announcements come as companies are going all in on AI agents to enhance their workflows. A 

 found that 82% of organizations plan to incorporate AI agents in their companies within one to three years.

AWS is taking the grunt work out of building and deploying AI agents

AgentCore Gateway, AgentCore Browser Tool, AgentCore Observability are some of the tools available in Amazon Bedrock AgentCore’s suite of services.

Data centers around the globe need to brace themselves for the 

 of the future—and the ones that are already here.

 by XDI, a physical climate risk data and analysis company, which examined the physical risk from different climate hazards for 8,868 data centers worldwide and deemed more than one-fifth of facilities (22.04%) at high or moderate risk. By 2050, XDI predicts that this proportion will increase to 26.73%.

 XDI’s findings are based on a “high-emission scenario,” a term used to describe a high-risk climate simulation where greenhouse gas emissions continue to increase with little intervention between now and the end of the century. The company describes high-risk facilities as those more likely to experience total or partial physical damage within their expected lifespan, and moderate-risk facilities as those that will experience higher insurance costs from climate hazard damage.

Examined hazards included forest fires, extreme wind, coastal inundation, tropical cyclone wind, freeze-thaw, soil movement, riverine flooding, and surface water flooding. Of the eight different climate change hazards, riverine flooding and coastal inundation (coastal flooding) were ranked as being very high levels of concern for data center infrastructure.

The findings come at a time when data centers have a reputation for causing environmental damage. While big companies like 

 have begun to experiment with unique facility designs intended to cut down their carbon emissions, the impact of AI-fueled power demands on the environment remains a concern. Morgan Stanley research 

 the data center industry will emit 2.5 billion tonnes of greenhouse gas emissions globally between 2024 and the end of the decade.

Drops of water on the outside of a server case.

Data centers have a climate problem on their hands

A recent report found that more than one-fifth of data centers worldwide were at high or moderate risk for climate hazard damage.

ITBR-Cloud migration strategies audio embed

Moving furniture into your friend’s new dorm usually involves beer and pizza as a reward. Let’s hope the IT team that helped Babson College’s big digital move in 2023 got enough slices.

Over the span of about six months that summer, the Massachusetts-based school, according to its CIO Patty Patria, migrated more than 20 terabytes of data for 6,000-plus campus employees. The emails and documents—and all the links and connections associated with them—went from multiple cloud environments, including Google Workspace, to one consolidated platform: Microsoft 365.

Patria and two IT pros who’ve conducted similar migrations shared with IT Brew how to prepare so data arrives safely at its new destination—and how to respond effectively if it doesn’t.

“Any time you change an interface that somebody’s been using for five, 10, 15, 25 years, that’s a challenging thing for them, and so you have to be very careful of planning,” Patria told us.

Software-as-a-service (SaaS) migrations are common during mergers and acquisitions, when the consolidating companies have clashing platforms. Babson, however, migrated data to reduce storage costs and improve collaboration between students and teachers, Patria said.

BetterCloud’s 2025 “State of SaaS Security,” which polled 568 US-based IT pros, found that the average company reduced their SaaS apps from 2022 to 2024 and orgs now have fewer SaaS applications than in 2021.

Digital storage can be a bit like your old closet: filled with years of stuff. Patria and her team monitored active and inactive users and made decisions regarding which data would make the trip—and how much mail to bring over, for example. “It’s really reviewing, if we haven’t accessed this data more than two years, three years, five years, do we need to keep this data?” Patria said.

Richard Harbridge, CTO at 2toLead, recommended companies perform a security and compliance audit of existing data to find sensitive, regulated content like PII, financial numbers, or health info across your SaaS repositories like Google Drive and Slack.

“If you find, for example, documents subject to 

, you’ll want to ensure equal or better protections in M365 or decide if that data should even move at all,” Harbridge wrote to us in a follow-up email.

A manual file copy from Google Drive to Microsoft’s OneDrive will not retain sharing permissions, version history, or external share links; migration tools (like ShareGate for migrating permissions) help with the effort, in theory. Some elements still must be reconfigured at the destination.

One of the toughest days for Patria and her staff, she said, occurred when documents arrived in unexpected condition. A third-party vendor had capabilities to migrate files over with unique permissions, Patria said, but the permissions were so unique that they didn’t carry over and required ongoing maintenance and discussions.

Each non-Microsoft SaaS has elements that don’t have a one-to-one counterpart in Microsoft 365. Google Forms, for example, cannot be auto-migrated and must be recreated in Microsoft Forms or exported to a PDF before transfer, Harbridge noted in his follow-up email, and Slack emoji reactions won’t always export into Teams. Email formatting may change, and connection points, like a Jira ticket pointing to a Google file, may break.

Mapping, however, leads to a possibility for upgrades and opportunity, Harbridge said, like upgrading to multi-factor authentication or activating data-governance features in, say, Microsoft Purview.

“It’s a great opportunity, since you have disruption, to change things and improve your security posture,” he told us.

Migrations can occur in phased content batches or a “big bang” option—one preferred by Will Thomas, managing director and cloud optimization lead at Protiviti—that stages and loads up 

content for migration while tests and pilots occur. Then the team can spend a weekend syncing a smaller workload: any updated, new, or removed data since that primary staging execution. That kind of “bang,” also known as a cutover, still requires “hypercare” to assist users with questions. Thomas said this support can include “ask me anything” sessions, or even a proactive, concierge-style treatment where the IT pro contacts users.

Harbridge, however, says there is no big bang: “Everything is phased and everything is scheduled.” When migrating file data from file shares to SharePoint, one might do an initial bulk transfer over a weekend, according to Harbridge, followed by incremental “delta” changes, which migrate any data that changed since the last transfer.

He recommends extra padded support hours for global users and the employees catching up after hours. “A lot of this stuff doesn’t happen during core time,” he said.

In a follow-up email to IT Brew, Patria said the IT team performed 65 training sessions, including office hours, and instructed a total of 96 people during the initial roll-out; she said the school offers ongoing training.

Babson users moved from Google Drive to Microsoft’s combination of OneDrive for personal data and SharePoint for shared documents.

Learning how to access documents in a new platform proved to be a tough challenge for some users, Patria said.

“Every time you do any one of these projects, as much change management and communication that you do, you could probably always do a little bit more,” Patria told us.

A big driver for SaaS migrations is that companies want to use AI—and they can’t as easily with data fragmented across legacy SaaS tools, according to Harbridge. Patria said the migration ultimately allowed them to standardize in Microsoft’s cloud and more easily work with Microsoft’s Copilot tool—another big move.

“I think cloud platforms set you up for greater transformations in other areas,” she said.

A weather cloud with three ethernet cables plugged into it

How to plan ahead for cloud migration frustrations

And how proactive decisions can cut down the time.

A cloud bill is likely not a one-page print. And if yours is, you may just be at your 

The cost-management homepage for Amazon Web Services features windows, bar graphs, recommended actions, summaries, and trends.

Such a data-filled view can be a “scary-looking dashboard,” according to George Crosby, founder of Cogito Group, a consultancy that tries to ease the fright and find ways for clients to cut cloud costs.

Crosby and Deloitte Principal Chris Thomas spoke to IT Brew about how to be ready for the cloud check when it arrives.

Cloud costs can include a variety of computing, network, and storage services. Crosby recommends searching a platform’s dashboard for top drivers and then drill into their causes.

A cost-management homepage prominently features monthly cost breakdowns, with colors for each service used: Say, the computing platform Amazon Elastic Compute Cloud (EC2), Amazon Relational Database Services, or application-deployment tool Red Hat OpenShift Service on AWS.

Let’s say EC2 costs are high. Crosby likes to use AWS’s filters in the “report parameters” panel to select EC2’s service and “usage type,” which reveals breakdowns for individual services. The EC2 usage-type list may reveal high “data transfer” costs.

When an availability-zone (AZ) outage occurs, an EC2 instance in that AZ shuts down. Some companies, as insurance for uptime, may have multiple instances and data transfers across both. A client should, in this case, investigate service-level agreements, to determine the need for the extra EC2 causing the data transfer, Crosby recommended.

Maybe the EC2 usage view reveals too many “on-demand” instances, and spot instances (which launch depending on available capacity) may be preferred—perhaps for build nodes that service internal users and not production environments, Crosby said. A migration to different CPU types, like ARM64 processors within cloud services, could also help to cut costs, he advised.

Crosby recommended an org use cloud-monitoring tool to reveal a 30-day and seven-day history and identify trends related to amounts of compute, storage, and data transfer.

“Have we over-engineered what’s currently running in the cloud? That’s one way of looking at it,” Crosby told us.

A 2024 State of Cloud report from IT expense manager Tangoe found that cloud software led to average business costs of $2,559 per employee annually after analysing 500 US finops practitioners. More than nine in 10 (92%) respondents reported that their cloud spending rose.

“This escalation is driven largely by the adoption of AI and GenAI technologies, likely because of the storage and processing requirements of large language models,” the report read.

Developers may be running expensive prompts in loops, or failing to shut down GPU-intensive training jobs, according to Thomas, Deloitte’s US hybrid cloud infrastructure leader.

Practices like usage quotas, automated alerts for runaway queries, and training programs can help address surprise costs, Thomas said.

And, when it comes to cloud resources, don’t rack up a big bill at lunch time.

“Sometimes [organizations] might run the cafeteria website on the most expensive tier of storage, for example, when really the lowest tier is all that’s needed and you need to save the highest tier for your business-centric workloads,” he told us.

A cloud service bill floating on top of an actual cloud

How to spot the cost-saving possibilities in your cloud bill

 Data transfers, on-demand services, AI, and other items to track monthly

Like Paris Hilton in 2003, Amazon Web Services (

) wants cybersecurity professionals to embrace the simple life.

AWS CISO Amy Herzog took the stage on day two of the company’s annual re:Inforce conference this week in Philadelphia to share some of the cloud giant’s latest security innovations, each aiming to simplify capabilities on existing tools. Herzog took on the role as CISO of the 

 earlier this month, replacing Chris Betz. She was previously CISO for Amazon’s AGI, ads, devices, and global media and entertainment units.

Herzog kicked off Tuesday’s keynote by announcing updates to AWS Identity and Access Management (IAM), the cloud giant’s web service that allows organizations to manage who has access to AWS resources. She said IAM, launched in 2011, was designed to “address the complex needs of modern cloud authentication and authorization at a massive scale,” and handles 1.2 billion API calls per second worldwide, she said.

“That means that 1.2 billion times per second, IAM is asked to determine if an API call should be permitted or denied,” she said.

Herzog announced that starting Tuesday, internal access findings will be made generally available as a new IAM Access Analyzer capability. The automated reasoning-powered tool analyzes policy types—including service control policies and resource control policies—to identify which roles or users have access to S3 buckets, Amazon DynamoDB tables, or Amazon Relational Database Service snapshots.

 Herzog also discussed new network protection capabilities. AWS customers will now be able to preview AWS Shield’s network security director. AWS Shield is a managed distributed denial of service protection service.

The new capability will allow AWS customers to easily pinpoint network security configuration issues by displaying identified risks and their severity level and providing specific remediation recommendations for them.

“It’s kind of like having a team of AWS security experts monitoring your network architecture and giving you the bottom line,” Herzog said.

Herzog also announced a new “simplified console experience” for AWS Web Application Firewall, which she said reduces the steps needed to configure initial application security by 80%. Rob Kennedy, AWS VP of network services, told IT Brew that the update simplifies the WAF security setup process for customers.

“Before there was a I would call it a complex process to go through, which was time-consuming…Today you can go in, you basically select the type of application that you want to protect, and straight away, [it] will give you a recommendation of what you should set up,” he said.

 During her keynote, Herzog touted some of AWS’s security feats occurring behind the scenes. She mentioned that AWS IAM now has a 100% 

 enforcement rate for root users across all AWS account types.

The freshly appointed CISO also highlighted the hard work of the cloud giant’s active defense tools.

“In the four months between December 9, 2024 to April 30 of this year, AWS prevented 943.6 million attempts to perform unauthorized encryption of customer S3 objects,” she said.

AWS CISO Amy Herzog on stage at re:Inforce.

AWS doubles down on IAM and network protection capabilities at re:Inforce

During her keynote, Herzog discussed several updates to AWS’s network protection and identity and access management capabilities.

The sky is the limit—unless you’re Microsoft, in which case, it is a booming business line that has solidified its place as a top cloud giant over the years.

When talking about cloud hyperscalers in the industry, Azure is a name that is hard to ignore. According to 

, Azure had a global market share of 22% in Q1 2025. In the same period, AWS had a global market share of 29%, while Google Cloud had 12%.

To understand how Azure secured a spot as one of the 

, IT Brew caught up with cloud computing experts to learn about some of the core decisions made in the cloud giant’s early days that positioned it for success.

Mark Russinovich, Azure CTO and technical fellow, told IT Brew that the story behind Azure began in 2006 when Dave Cutler, known for his key role in the development of Windows NT, set off on a project to create a platform for the emerging “software-plus-service era.” The internal project was known as Project Red Dog.

“This team with Dave, under [then Microsoft Chief Software Architect] Ray Ozzie, started to look around the company and talk to different teams that had services, including teams like Hotmail and the Windows Live team, about what they needed to make it easier for them to run services,” Russinovich said.

Azure, formerly Windows Azure, was officially announced in 2008 at the Professional Developers Conference and made its commercial debut in 2010. Brand-naming agency Lexicon claims to have come up with the name Azure, which means sky blue, to reflect Microsoft’s “new attitude” and the “open-ended and flexible nature” of the platform.

“As far as the name Azure, marketing picked that,” Russinovich said. “I thought it was an odd pick at the time because Azure is the color of the sky with no clouds in it, which is the opposite of cloud.”

Wade Wegner, a director of technical evangelism for Azure from 2010 to 2012, told IT Brew that Azure was immediately attractive to Microsoft’s “loyal base,” which included .NET developers. However, outside of that, things looked slightly different, partly because of Amazon Web Services’s (AWS) 

“The broader developer ecosystem actually was centered and flocking around what AWS was doing and it was hard for us, especially early days, given we had such a very specific platform as a service approach to building for the cloud,” Wegner, now a chief ecosystem and growth officer at DigitalOcean, said.

Karthik Ranganathan, founder and co-CEO of Yugabyte, an open-source distributed SQL database company, added that Microsoft’s later entrance into the cloud market ultimately became a focal point for industry onlookers in Azure’s early days. Ranganathan worked as a software design engineer at Microsoft between 2005 and 2007.

“The critics would point out that this gap is going to be very tough to bridge because Amazon’s running away with it, or so to speak,” he said.

Still, several decisions made along the years helped propel the cloud giant as a serious contender in the cloud race. For starters, Russinovich said the company’s move to release Azure Virtual Machines, an infrastructure-as-a-service (IaaS) offering, in 2012 “unlocked a lot of business.”

“What we found was, there were customers that wanted that kind of convenience of what the platform could do for them but a lot of enterprises had existing software that they wanted to just migrate into a cloud platform, and not having an IaaS capability meant that they couldn’t use Azure,” Russinovich said.

Russinovich added that Microsoft’s growing support of open-source, 2018 acquisition of 

, and strong relationship with enterprises also shaped Azure’s trajectory.

“We had this long history of relationships and understanding what enterprises wanted,” Russinovich said. “So, we could bring that to the design of our platform from a security, reliability, operation, [and] support perspective.”

Others, like Cockroach Labs Senior Director of Global Alliances Cassie Zimmerman, credit current Microsoft CEO Satya Nadella, who took on the role in 2014, as another big driver behind Azure’s trajectory. (Cockroach Labs is an Azure database partner.)

“He was really the one who came in and declared Microsoft as this cloud-first, mobile-first company,” Zimmerman said. She pointed to Azure’s rebrand from Windows Azure to Microsoft Azure shortly after Nadella took the reins as a reflection of that shift.

Today, Russinovich said Azure is in the “AI era.” Recent decisions at the company actively reflect this. Microsoft has partnered with OpenAI since 2019 and has pumped 

, the tech company announced that it recently added xAI’s Grok 3 models to the Azure AI Foundry, along with several updates to the unified application platform.

Ranganathan said Microsoft’s early wager on AI has ultimately paid off.

“I’m sure they diversified into a lot of bets and this was the bet that really started taking off in the market,” Ranganathan said.

Beyond AI, Ranganathan said industry onlookers may be waiting for the cloud giant to “round out” its offering for those that fall outside of its enterprise audience.

“Its origins were so much in the enterprise stack…that some of the workflows and some of the ways that developers consume it, it’s complicated,” Ranganathan said. “It works great for large enterprises. It’s difficult for a startup sometimes.”

A Microsoft building peeking through trees.

How Azure rose to the ranks as a cloud giant

“I thought it was an odd pick at the time because Azure is the color of the sky with no clouds in it, which is the opposite of cloud,” Microsoft Azure CTO and technical fellow tells IT Brew.

It’s almost like Hobbs and Shaw—two heavyweights putting rivalry aside and teaming up against a common enemy: computing power.

, OpenAI plans to add Alphabet’s Google Cloud service to its list of compute providers. The cloud support will assist OpenAI in training its AI models and meeting the massive computational demands for its jobs, sources told Reuters.

The move raised eyebrows from analysts, given the two companies’ competition.

“Ultimately, we view this as a big win for Google’s cloud unit, but...there are continued worries that ChatGPT is becoming an incrementally larger threat to Google’s search dominance,” Scotiabank examiners wrote in a Tuesday note.

 In addition to search, both companies compete as AI builders. Alphabet’s Google DeepMind lab has created models like Gemini 2.5, while OpenAI has introduced recent LLMs like OpenAI o3 and o4-mini.

 $10 billion in annual recurring revenue, less than three years after launching ChatGPT in November 2022. The popular tool (which recently 

 as of March, according to the company (along with three million 

The company’s CEO Sam Altman has been open(AI) about the company’s need for power. “I think compute is going to be the currency of the future,” he said on the 

 one year ago. “I think it’ll be the most precious commodity in the world.” (He also tweeted in March that the company’s GenAI tool was 

The Google and OpenAI partnership—finalized in May—adds to recent deals for OpenAI to gather computing capacity. The company signed a 

 with cloud startup CoreWeave in May, and announced 

 a $500 billion investment in the AI infrastructure-building Stargate project.

Alphabet has been enthusiastic about its Google Cloud usage and growth. In its most recent earnings call, the company highlighted its GPU offerings, including Nvidia’s B200 and GB200 Blackwell chips, and its custom-designed tensor processing units (

). The Google Cloud Next conference in April showcased many business uses, like the introduction of AI security-focused agents for Dunn and Bradstreet, Google shared on its blog, and on-premise Gemini models for Nvidia and Dell.

According to Alphabet’s Q1 earnings report, Google Cloud revenue increased by 28% year over year to $12.3 billion.

The OpenAI and Google deal surprised one Gartner analyst—only because it took so long.

“I anticipate that OpenAI and Google and other frontier model innovators will continue innovating and exploring frontiers of infrastructures and model capabilities. Today the needs overlap, hence the mutually beneficial partnership,” Chirag Dekate, VP analyst at Gartner, wrote in an email to IT Brew.

He warned, however, Google now has to balance TPU usage among its cloud clients and internal projects like DeepMind. An agreement with a growing AI giant like OpenAI complicates resource planning.

“GPU leaders must redouble their efforts to lower cost structures for clients,” Dekate wrote.

Neither OpenAI nor Google responded to June 10 requests for comment about specifics of the partnership and how the partnership will benefit their respective companies.

OpenAI and GoogleAI logos on podiums surrounded by binary code.