11.26.2025

ForwardThinking 🤔

Turkey Day tech thoughts

Some readers may recall that around the July 4th holiday, rather than do my normal column where I take a deep dive into a subject that’s been on my mind, I instead paid homage to Dan Shaughnessy, the long-time sports columnist at the Boston Globe, who regularly writes columns with his random thoughts about the goings-on in sports news. With another holiday upon us, I decided to revisit the idea.

So here goes: Picked up enterprise tech news pieces while zipping through a short week and waiting for Turkey Day, time with family, feasting and football.

*It struck me kinda funny, struck me kinda sad when I read Satya Nadella’s benevolent, kumbaya piece about our AI future on LinkedIn recently. The title alone brought forth visions of rainbows and unicorns: A Positive Sum Future, Nadella’s idealistic view of our AI future as he steps back from his day-to-day CEO role and responsibilities to hang with the technical teams more. As Dan would say, swell.

He put forth this lovely thought: “Let us move beyond zero-sum thinking and the winner-take-all hype and focus instead on building broad capabilities that harness the power of this technology to achieve local success in each firm, which then leads to broad economic growth and societal benefits.” It sounds so great, it almost makes you forget he works for a company with a $3.5 trillion market cap, one that clearly hopes to dominate the AI market. I guess both things can be true.

Photo by Jiaqian AirplaneFan, CC BY 3.0 via Wikimedia Commons

It reminded me of another comment Nadella made in 2015 when he appeared on stage at Dreamforce with Marc Benioff, ending years of litigation and general hostility between the two companies. On that day he said, “It is incumbent upon us, especially those of us who are platform vendors to partner broadly to solve real pain points our customers have." It was all about working together, except when it wasn’t, at which point he would compete hard — and compete hard he has.

There’s nothing wrong with that of course. It’s what companies do, and when done well customers can benefit from the ensuing innovation. Just don’t pretend it’s anything more than that or it’s all about everybody winning because it rarely is.

Quiz (answer below): Name the company that is fourth in the cloud infrastructure market behind Amazon, Microsoft and Google.

*Some VC firms must be countin’ on a miracle when it comes to investing in AI startups, if the latest data from PitchBook is any indication. The firm found that investors plowed almost $55 billion into so-called AI startups in Q3, up from $49 billion the prior quarter. Four deals accounted for 46% of that amount: Anthropic scoring $13 billion, xAI landing $10 billion, Mistral pulling in $1.5 billion and finally nScale nabbing $1.1 billion — for a total of $25.6 billion for those four deals.

That’s a lot of capital concentrated in four companies, and it left the remaining 1,082 startups to split $29 billion or an average of around $27 million each.

A donut chart shows that four mega-deals dominated Q3 2025 AI VC funding, with Anthropic at $13B, xAI at $10B, Mistral at $1.5B, and Nscale at $1.1B, together totaling 46% of the $54.8B invested; the remaining 1,082 deals account for $29.2B.

Chart courtesy of PitchBook

There’s a strong belief that investors can win by investing in the biggest large language models because these will be the platforms where future software gets built. That’s probably a good bet, but there’s also a chance that these models become commodified over time and have less value, while costing billions and billions to operate. It’s clear that not everyone can win and some investors will get lost along the way.

*It seems that for Jeff Bezos, being founder and chairman of Amazon, a space company and owner of assorted yachts and a major newspaper isn’t enough. He needs a new toy to keep his head expanding, and he recently announced that he has a shiny new AI startup. He put up some of the $6.2 billion in initial funding, and named himself co-chief executive. Hard to believe he’s sharing the title, but it’s probably because he’s a little busy with his other responsibilities to devote full-time to his newest venture.

According to the NYT, the company is known as Project Prometheus and will focus on AI engineering with a manufacturing bent, specifically computers, cars and spacecraft. You can see how this might fit in with his other ventures.

The company, which has mostly stayed under the radar until recently, already has 100 employees. It will be interesting to see if Bezos has the Midas touch more than once or if billionaire impatience gets in the way.

On that note, go off and enjoy the holiday, be safe and we will be back next week reporting from AWS re:Invent in Las Vegas.

Quiz Answer: Alibaba with around 4% market share, per Synergy Research.

~Ron

What's new on the blog 📰

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+

Harness CEO Jyoti Bansal’s love affair with building startups

Harness CEO Jyoti Bansal once told me that some people have hobbies like running or hiking. He likes building startups.

I spoke to him recently at a company event in Boston. Turns out that he still likes building startups, and satisfies that desire inside his company by using what he calls 'a startup-within-a-startup approach.'

"So in this startup-within-a-startup model, you can innovate at low cost and still take plenty of risks," Bansal said.

Read full article>>

News of the Week 📣

Model updates from anthropic and gemini have people buzzing

Regardless of the time of the year, or the upcoming holiday, the onslaught of AI news continues unabated. So it should come as no surprise that neither Google nor Anthropic let the holidays stop them when it came to announcing major updates to their models, Gemini and Claude Opus respectively.

Google was first out of the gate with Gemini 3 and has been showered with praise about its speed and accuracy. It's only been out a short time, but as a regular user of Gemini, I have to say, I don't see much difference between the current version and the older one. It seems to me in terms of speed that it actually feels a bit slower while it deliberates and works through what the answer to your query is going to be (but I did generate the image below with the model's image generator).

Two side-by-side painted canvases on an easel show the logos and names of Claude Opus 4.5 on a cool blue-purple background and Google Gemini 3.0 on a warm yellow-orange background, with art supplies and brushes visible in a bright studio behind them.

AI-generated image created with Gemini Nano Banana Pro

Meanwhile Anthropic released Claude Opus 4.5, its AI coding-focused model, with its own hype where it promised faster and smarter, but perhaps more importantly cheaper paths to coding and agent development. In fact, The New Stack's Frederic Lardinois wrote on LinkedIn, API pricing has dropped to $5/$25 per million tokens, down from $15/$75. As he points out, that is significantly cheaper for far better performance. Faster and cheaper is definitely a combo enterprise buyers like.

It's important to note that regardless of the news surrounding these model releases and whatever benchmarks they have surpassed, it will always come down to real-world experience, and more importantly individual use cases. It's how it works for you that matters.

Amazon announces $50B HPC/AI data center investment with u.s. government

Not to be left out of the gaudy data center deals OpenAI and Anthropic have been announcing this year, Amazon announced a mega deal of its own this week with the U.S. government around high performance computing and AI. The deal could total $50 billion if it reaches its full value and involve 1.3 gigawatts of capacity. Construction is expected to begin next year.

"We're giving [government] agencies expanded access to advanced AI capabilities that will enable them to accelerate critical missions from cybersecurity to drug discovery. This investment removes the technology barriers that have held government back and further positions America to lead in the AI era," Amazon CEO Matt Garman said in a blog post announcing the deal.

Photo by Ron Miller

This makes the $10 billion JEDI cloud deal from 2018 look paltry. That deal, which was a winner-take-all affair, was eventually won by Microsoft, but litigation and other delays eventually led to the Department of Defense looking at other options. Amazon officials attributed the loss at least partly to political favoritism at the time.

Whether this deal will succumb eventually to similar political shenanigans remains to be seen, but for now, Amazon has itself a heck of a deal, especially if it reaches its full value.

a possible memory chip shortage in 2026 could derail some hyperscaler data center plans

A growing body of evidence suggests that we could be looking at a memory chip shortage in 2026, a result that could significantly slow hyperscaler data center expansion plans.

As companies line up for the increasingly precious resource, suppliers are having a hard time keeping up with the massive demand, according to multiple reports. There's not only the regular requirements of PC and laptop makers, but AI requires a significant amount of specialized memory, and that is what is putting tremendous pressure on the memory supply.

Close-up view of several memory modules laid out on a dark surface, with blue and green circuit boards and black chips in focus, highlighting computer RAM components.

Photo by Curated Lifestyle for Unsplash+

Nikkei Asia reports that memory maker SK Hynix, a South Korean memory chipmaker, is seeing customer concern about a shortage, perhaps fueled by companies like OpenAI, which reportedly signed a deal to procure 900,000 high-bandwidth memory (HBM) wafers per month next year to fuel Project Stargate — a figure that, if accurate, would represent a staggering share of global HBM output. A deal like that could suck up a lot of available inventory, leaving some customers wanting.

Whether this cascades across the market remains to be seen, but the tea leaves suggest that a shortage could have an impact on data center construction as companies like OpenAI clearly try to get ahead of any potential trouble.

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

The New Stack Makers

Amazon CTO Werner Vogels' Predictions for 2026

Frederic Lardinois

Axios

Deepfakes flood retailers ahead of peak holiday shopping

Sam Sabin

HFS Research

AI Won't Save Enterprises from Tech Debt Unless They Change the Architecture First

Hansa Iyengar & Phil Fersht

Platformer

Roblox is a problem — but it’s a symptom of something worse

Casey Newton

Look who's talking 👄

"It's so interesting because, especially with AI, more than any other technology, this is the one where I want startups to knock on my door"

- TD Synnex CIO Kristie Grinnell in FastForward profile this week.

Please subscribe to the newsletter and feel free to share.

3350 Virginia St, Suite 262, Miami, FL 33133

ForwardThinking 🤔

Turkey Day tech thoughts

What's new on the blog 📰

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

Harness CEO Jyoti Bansal’s love affair with building startups

News of the Week 📣

Model updates from anthropic and gemini have people buzzing

Amazon announces $50B HPC/AI data center investment with u.s. government

a possible memory chip shortage in 2026 could derail some hyperscaler data center plans

My adventures in AI 🤖

What I'm reading 📚

The New Stack Makers

Amazon CTO Werner Vogels' Predictions for 2026

Axios

Deepfakes flood retailers ahead of peak holiday shopping

HFS Research

AI Won't Save Enterprises from Tech Debt Unless They Change the Architecture First

Platformer

Roblox is a problem — but it’s a symptom of something worse

Look who's talking 👄

Subscribe

FastForward #48: Turkey Day Tech Thoughts

You may also like...