12.06.2025

ForwardThinking 🤔

AWS takes an AI step forward

A year ago this week, I launched FastForward #1 at AWS re:Invent. It was a heck of a place to officially start my journey, amidst the tumult of a ginormous conference in one of the most distracting places on the planet, but we were ready to go and there is never a perfect time. So I took the leap.

A year and 48 issues later, I was back in Vegas waiting in long taxi lines and getting hassled by overzealous event security, who took umbrage with me for having the audacity to take a picture on my way out of the auditorium during Matt Garman’s keynote. I was threatened with arrest and having my badge revoked until cooler heads prevailed. I was sent on my way, shaken up and aggravated, but with a good story to tell my journalist friends in the press room.

AWS CEO Matt Garman delivering keynote at AWS re:invent 2025 in Las Vegas.

AWS CEO Matt Garman delivering the keynote at re:Invent 2025. Photo courtesy of Amazon.

But the real story this week wasn’t about my misadventures one year into FastForward. It was about the themes of re:Invent and where Amazon stands in the ongoing AI race. Two years ago Adam Selipsky was still AWS CEO and there was a general industry perception that Amazon had fallen behind. Microsoft’s tight relationship with OpenAI at the time seemed to give Redmond a leg up. Everyone said it was early, but some level of panic seemed to be setting in at Amazon.

Six months later, Selipsky was out (officially stepping away), replaced by AWS lifer Matt Garman, who set the company more firmly on its AI journey. Last year Garman’s inaugural keynote emphasized the basics, Amazon’s infrastructure strengths, before launching into the AI announcements. This year agentic took center stage with a much more coherent AI story that included chips, hardware, agent creation tools and Amazon large language models. But was the company losing something by avoiding the core infrastructure story?

Where to focus?

If perception is everything, then Amazon covered its bases with the flurry of announcements this week in Las Vegas. “It certainly helped them catch up in all the key areas,” Holger Mueller, an analyst at Constellation Research, told FastForward this week. He particularly liked things like larger S3 buckets, which are now able to handle objects up to 50 terabytes, large enough to hold most SAP ERP data. He also highlighted the release of S3 Vectors, which lets customers query vector data at what Amazon claims is a fraction of the cost of stand-alone vector databases. Finally, he liked the agent story, including the Amazon Bedrock AgentCore update and the announcement of new all-purpose agents for developers, security and DevOps.

But David Linthicum, founder and principal analyst at Linthicum Research, offered a pointed critique of Amazon's AI focus. “My overall take is that these announcements feel more like “me-too” moves than true innovation. In the rush to chase generative AI hype, AWS risks sidelining the mission-critical services enterprises depend on,” Linthicum said. “In short, AWS is back in the AI headlines, but these announcements don’t address the everyday operational challenges that keep enterprises up at night.”

"In the rush to chase generative AI hype, AWS risks sidelining the mission-critical services enterprises depend on."
~David Linthicum, cloud consultant

As Jon Turow, a partner at Madrona Venture Group and a former AWS employee, told me, he’s not completely sold on Amazon's approach." The traditional AWS Lego brick approach may not be the right abstraction level for this generation of technology,” Turow said, referring specifically to AgentCore. “Developers are discovering other intermediate levels of abstraction for agentic infrastructure that are higher level while still being composable." Those intermediate abstraction levels hide some of the complexity of building agents, whereas the Amazon approach provides more basic, but flexible tooling.

Amazon may still have miles to go, but at least it finally took the leap — much like I did last year when I published the first issue of this newsletter. You can hardly blame the company for sticking with familiar approaches as it more fully incorporates AI into the product set. The real test is whether it can push ahead on agents without losing sight of the infrastructure foundation that has made AWS what it is.

~Ron

What's new on the blog 📰

Amazon Bedrock AgentCore getting key governance updates

I was at AWS re:Invent this week in Las Vegas where I managed to learn some things including AgentCore's new governance features, pieces that are absolutely essential for controlling and understanding agent behavior.

Read full article>>

Diagram showing Amazon Bedrock AgentCore at the center of a hexagon, with six surrounding icons representing core agent components: security, code, search, time/scheduling, data storage and system operations.

Image courtesy of Amazon

The new CIO asked one crucial question that helped refocus this Fortune 100 company

When Kristie Grinnell joined TD Synnex at the beginning of this year, she started the process of getting to know the large company by taking what she called a listening tour. And she broke the ice with one simple question:

“If I had a magic wand and I could grant you a wish, what would that wish be?” It quickly exposed the real friction points and what employees wished they could change.

Read full article>>

Microsoft, Google chip away at Amazon’s cloud dominance as neoclouds rise

The cloud market is shifting ever so slightly, as Amazon's total market share dominance has started showing some cracks. AWS is feeling more pressure from Microsoft and Google, and we’re also seeing explosive growth from neocloud providers at the lower end of the market, where even small gains can still translate into big bucks.

Read full article>>

A glowing cloud icon sits on top of a dark circuit board surface, with streams of binary code flowing into and out of the cloud, suggesting cloud computing and data movement.

Photo by Getty Images on Unsplash+

How Microsoft tests its own developer tools to build an AI-first engineering organization

If Microsoft wants to sell AI-coding tools, the best way to do that is to test them internally before they go out to customers. That not only pressure-tests these tools under real-world enterprise conditions, it helps transform Microsoft's engineering approach in the process.

I spoke to Amanda Silver, CVP and head of product for apps and agents to discuss the company's approach and how it's trying to remove the grunt work from the development process.

"Our focus now is on getting the most miserable, soul-draining parts of the job and transforming them with AI, as opposed to just generating AI code," she said.

Read full article>>

Red Hat’s CTO sees AI as next step for company’s open approach

I recently spoke to Red Hat CTO Chris Wright, who sees AI as a logical extension of what Red Hat has been doing for years around Linux, hybrid cloud and containerization. For him, inferencing is the new runtime:

"Our view is that inferencing is the production runtime for AI. A model doesn't do something useful for you until it’s got an API and it’s serving content. That content is served through inferencing," he said.

Read full article>>

Celonis wants to be the roadmap to rebuild workflows for agentic AI

If agentic AI is giving companies the opportunity to redesign their workflows, this has to be more than an intellectual exercise. It requires tools to map the complex way work flows and interconnects across a company and its internal systems.

That's precisely what Celonis, a German startup launched in 2011, is attempting to do, and it suddenly finds itself in the right place at the right time as companies look to rethink their processes in an agentic context.

Read full article>>

Aerial view of a highway splitting into two directions through a dense green forest, symbolizing a crossroads or decision point.

Photo by Unsplash Community on Unsplash+

News of the Week 📣

Salesforce has decent quarter, but Wall Street still not convinced

Salesforce had a decent quarterly earnings report this week reporting $10.3 billion in total revenue, up 9% over the prior year. While the company slightly missed analysts revenue predictions ($10.26 billion vs. $10.27 billion expected), it handily beat earnings per share predictions ($3.25 adjusted vs. $2.86 expected). These numbers are per CNBC.

CEO Marc Benioff was typically excited about the results, attributing the growth to increasing interest in the company's agentic AI product Agentforce. He claimed the company now has 6000 paying customers using Agentforce. But in spite of the good financial news, the stock is down 30% this year, and is selling at a level that Yahoo Finance referred to as "historically cheap."

Salesforce CEO Marc Benioff speaking animatedly at Dreamforce 2024.

Photo courtesy of Salesforce

That could be because investors are anticipating the possible disruption of SaaS platforms like Salesforce by agents, something that is being hotly debated in the industry right now. But Benioff sees AI extending SaaS for those who are taking advantage of the opportunity. "It's not about the fundamental, I would say, elimination of SaaS. What I would say, it's the fundamental extension of SaaS," Benioff said.

The revenue train is expected to keep rolling as the company also upgraded its revenue guidance for the next fiscal year. Robin Washington, president and chief operating & financial officer had this to say at the earnings call: "We are pleased to raise the low end of our fiscal year '26 revenue guidance to $41.1 billion to $41.3 billion," she said.

Amazon and google announce service to make it easier to move data between rival clouds

Last year, the big three cloud providers introduced programs waiving some data egress fees for customers migrating off their platforms, reducing the cost of moving data out of a given cloud. This week, Amazon Web Services and Google Cloud each announced a multicloud networking service to move data at high speed between the two rivals. These offerings provide private, high-bandwidth connections between the platforms and should make it easier to shift data than relying on general-purpose, internet-based migration tools.

"Previously, when interconnecting workloads across multiple cloud providers, customers had to go the route of a ‘do-it-yourself’ multicloud approach, leading to complexities of managing global multi-layered networks at scale," the company wrote in a press announcement.

Stylized 3D illustration of a blue cloud icon with three arrows pointing downward to groups of abstract user figures, representing cloud computing distributing resources to multiple users.

Photo by Alex Shuper on Unsplash+

The product called AWS Interconnect-multicloud by Amazon and Cross-cloud Interconnect by Google is designed to eliminate the DIY aspect of moving between clouds by providing a product that is specifically built for this purpose.

One of the big areas of friction for large enterprises operating in the cloud has been moving data between different clouds. It has historically been an expensive and difficult endeavor, essentially locking a company into whatever cloud vendor they were using for a particular workload. This week's announcements are a step toward making it easier for enterprises to have true multi-cloud strategies.

Openai Blames customer profile data leak on Mixpanel

Perhaps lost in the Thanksgiving holiday shuffle and the ongoing onslaught of news, OpenAI recently announced a data leak involving its third-party data analytics provider, Mixpanel.

The company was careful to point out that no OpenAI data was directly affected by the breach. "This was not a breach of OpenAI’s systems. No chat, API requests, API usage data, passwords, credentials, API keys, payment details, or government IDs were compromised or exposed," the company wrote in a blog post announcing the leak. But user profile data related to the use of the API product, which was stored and maintained by Mixpanel might have been compromised, according to the company.

Purple OpenAI logos floating on a purple and black striped background.

Photo by Mariia Shalabaieva on Unsplash

The company dropped Mixpanel as a result of the incident. "As part of our security investigation, we removed Mixpanel from our production services, reviewed the affected datasets, and are working closely with Mixpanel and other partners to fully understand the incident and its scope," it wrote.

It seems parting ways with the vendor was a way to show users that a leak like this isn't acceptable, but it's also unfortunately a common occurrence, no matter how careful vendors try to be.

My adventures in AI 🤖

Photo by Andy Kelly on Unsplash

I've been thinking a lot about the hype around the Gemini update this morning, especially the speed increase I'm hearing about. I haven't seen that yet in my usage. In fact, if anything, it seems to work more deliberately than ever, which may be a good thing if it means more accurate responses.

But whatever advantage Gemini may or may not have, I find this hype cycle around each model update to be a bit exhausting. Can we all just agree that at some point if this advantage is real, other model makers will catch up and maybe surpass it in future versions and the cycle will begin anew.

Nobody is going to take this market by storm and pull ahead for good. It's just the nature of this game.

This post originally appeared on my LinkedIn.

What I'm reading 📚

Person reading book, but all you can see is their hands holding the book. They are sitting on a hardwood floor steeped in shadow, creating an interesting look.

Photo by Blaz Photo on Unsplash

CNBC

Amazon Nova Forge lets clients customize AI models for $100,000 a year

Jordan Novet

FT

Oracle is already underwater on its ‘astonishing’ $300bn OpenAI deal

Bryce Elder

Bloomberg

The Rise of AI Reasoning Models Comes With a Big Energy Tradeoff

Rachel Metz & Dina Bass

TechCrunch

What is Bending Spoons? Everything to know about Eventbrite’s acquirer

Anna Heim

Look who's talking 👄

"In short, AWS is back in the AI headlines, but these announcements don’t address the everyday operational challenges that keep enterprises up at night."

- David Linthicum, founder and principal analyst at Linthicum Research in FastForward this week.

Please subscribe to the newsletter and feel free to share.

3350 Virginia St, Suite 262, Miami, FL 33133

ForwardThinking 🤔

AWS takes an AI step forward

Where to focus?

What's new on the blog 📰

Amazon Bedrock AgentCore getting key governance updates

The new CIO asked one crucial question that helped refocus this Fortune 100 company

Microsoft, Google chip away at Amazon’s cloud dominance as neoclouds rise

How Microsoft tests its own developer tools to build an AI-first engineering organization

Red Hat’s CTO sees AI as next step for company’s open approach

Celonis wants to be the roadmap to rebuild workflows for agentic AI

News of the Week 📣

Salesforce has decent quarter, but Wall Street still not convinced

Amazon and google announce service to make it easier to move data between rival clouds

Openai Blames customer profile data leak on Mixpanel

My adventures in AI 🤖

What I'm reading 📚

CNBC

Amazon Nova Forge lets clients customize AI models for $100,000 a year

FT

Oracle is already underwater on its ‘astonishing’ $300bn OpenAI deal

Bloomberg

The Rise of AI Reasoning Models Comes With a Big Energy Tradeoff

TechCrunch

What is Bending Spoons? Everything to know about Eventbrite’s acquirer

Look who's talking 👄

Subscribe

FastForward #49: AWS takes an AI step forward

You may also like...