Newsletter
AI

My AI Diary: November 17 — The AI Race Is No Longer Just About Smarter Models

Photo: Levart_Photographer on Unsplash
In this article 12 sections
  1. By Thomas | financial enthusiast
  2. 1. Microsoft Is Tired of Depending on Everyone Else
  3. 2. Anthropic’s IPO Could Become the Ultimate AI Reality Check
  4. OpenAI and NATO: AI Security Is Becoming Critical Infrastructure
  5. 4. The AI Safety Debate Just Became More Serious
  6. 5. The Real AI Bottleneck Is Still Money, Power and Infrastructure
  7. What This Means for My Watchlist
  8. First, who controls the infrastructure?
  9. Second, who can deliver AI at a sustainable cost?
  10. Third, who gets paid when AI becomes essential?
  11. My Take: The AI Industry Is Growing Up
  12. Sources and Further Reading

Microsoft challenges OpenAI, Anthropic heads toward Wall Street, and AI security becomes a serious business. Today made one thing clear: the next phase of artificial intelligence is going to be expensive.

By Thomas | financial enthusiast

November 17, 2026 · 5 min read


I started going through today’s AI headlines expecting the usual competition over benchmarks, model sizes and impressive demos.

Instead, I found myself looking at something much bigger.

Microsoft wants to build more of its own AI technology. Anthropic is preparing for a massive IPO Initial public offering. A company selling its shares on a stock exchange for the first time.. OpenAI is expanding into NATO cybersecurity. And regulators are increasingly asking uncomfortable questions about what happens when autonomous AI systems go wrong.

That is quite a lot for one Tuesday.

But what caught my attention wasn’t any single announcement. It was how everything seems to connect.

AI is moving from a technology race into an infrastructure, security and profitability race.

And I think investors need to pay attention.

1. Microsoft Is Tired of Depending on Everyone Else

One of today’s biggest announcements came from Microsoft.

At its Ignite conference, the company introduced a new collection of specialized AI models, including MAI Code 2 for programming and a separate model focused on cybersecurity.

Microsoft says its coding model can approach the performance of competing systems while operating at significantly lower cost.

That last part matters.

For years, the conversation has been about who can build the smartest model. But intelligence alone doesn’t necessarily create a profitable business.

If Microsoft can deliver similar results for less money, that changes the economics.

And there is another interesting angle here.

Microsoft has invested heavily in OpenAI. Now it is increasingly developing its own alternatives.

I don’t see this as an immediate breakup. I see it as a company making sure it controls more of its technological future.

From an investor’s perspective, that makes sense.

Why depend entirely on another company for technology that could become central to your entire business?

What I’m watching: Whether Microsoft’s own models can reduce the cost of delivering AI through Azure, GitHub and its enterprise products.

Because cheaper AI could eventually matter more than another impressive benchmark.

2. Anthropic’s IPO Could Become the Ultimate AI Reality Check

The Anthropic story is fascinating.

The company is moving forward with plans for a public offering that could raise more than $60 billion.

That’s an extraordinary amount of money.

And the timing is interesting.

AI valuations have become increasingly controversial. Infrastructure costs are enormous. Investors are questioning how quickly companies can turn AI adoption into sustainable profits.

Yet Anthropic is pushing ahead.

According to reporting from the Associated Press, the company has substantial long-term computing commitments and remains unprofitable under conventional accounting measures.

This is where I start thinking less like a technology enthusiast and more like an investor.

A brilliant AI model is one thing.

A business that can finance its infrastructure, retain customers, defend its margins and generate Free cash flow Cash a company generates after paying for operations and investments. It can fund dividends, buybacks or debt repayment. is something completely different.

Anthropic may eventually prove that the economics work.

But the IPO could force investors to answer questions that private-market valuations have allowed them to postpone.

How much revenue can frontier AI companies realistically generate?

How much will computing costs decline?

And how much of the future profit will actually remain with the model developers rather than flowing to chipmakers, cloud providers and electricity suppliers?

I keep coming back to that last question.

What I’m watching: Anthropic’s valuation relative to revenue, cash burn and long-term infrastructure commitments.

The offering could become one of the most important tests of investor confidence in the AI industry.

3.

OpenAI and NATO: AI Security Is Becoming Critical Infrastructure

Another development that stood out today was OpenAI’s expanding role in cybersecurity.

The Wall Street Journal reported that NATO’s Communications and Information Agency is using OpenAI models for authorized cybersecurity work.

The applications include investigating suspicious activity, identifying vulnerabilities and helping security teams respond to threats.

This feels like an important shift.

We have spent years discussing AI assistants, chatbots and productivity tools.

Now the technology is moving deeper into national security and critical infrastructure.

And that creates both opportunities and risks.

On the positive side, AI could help security teams analyze threats faster than human analysts working alone.

But the same technological capabilities can potentially be misused.

An AI system capable of finding vulnerabilities is valuable to defenders. It can also be dangerous if the wrong people gain access to its capabilities.

That makes safeguards, access controls and oversight increasingly important.

For investors, cybersecurity may become one of the more durable commercial applications of AI.

Not because the technology is exciting, but because the underlying problem isn’t going away.

Organizations need to defend their systems regardless of whether the broader AI market is rising or falling.

What I’m watching: Whether AI security products become recurring, profitable enterprise services rather than expensive demonstrations.

4. The AI Safety Debate Just Became More Serious

Today’s headlines also highlighted the growing legal risks surrounding autonomous AI agent An AI system that plans and carries out multi-step tasks on its own, using tools such as a browser or code..

California’s attorney general has sued OpenAI over allegations related to inadequate safeguards and an earlier incident involving the Hugging Face platform.

The allegations raise a question that I think the industry still hasn’t fully answered.

What happens when an AI agent takes an action nobody intended?

Traditional chatbots mostly generate information.

Autonomous agents can interact with websites, execute code, access external services and potentially modify real systems.

That’s a different risk category.

And it makes the discussion about AI safety much more practical.

Companies need systems that can monitor agents, restrict permissions, record actions and intervene when something goes wrong.

I suspect this will become a major enterprise requirement.

The more autonomy businesses give AI systems, the more important control mechanisms become.

For me, the interesting opportunity isn’t only in building more capable agents.

It’s also in building the tools that make those agents safe enough to deploy.

5. The Real AI Bottleneck Is Still Money, Power and Infrastructure

After reading through today’s developments, I ended up returning to a familiar theme.

Infrastructure.

Microsoft needs computing capacity for its models.

Anthropic needs enormous amounts of capital to finance its ambitions.

OpenAI needs reliable infrastructure for increasingly sensitive applications.

None of these businesses can operate at scale without data centers, chips, networking, memory, cooling and electricity.

And that is why I continue watching the companies supplying the physical foundations of AI.

Nvidia NVDA ▼2.9% remains important, obviously.

But the opportunity is much broader than GPU Graphics processing unit. A chip with thousands of small cores; the workhorse for training and running AI models..

Power generation, electrical equipment, cooling technology, memory and networking could all benefit from continued AI infrastructure spending.

There is a catch, though.

Higher spending doesn’t automatically mean higher profits.

If AI companies build capacity faster than customers generate economic value from it, the industry could face a painful adjustment.

We’ve seen similar patterns in previous technology investment cycles.

The infrastructure can be genuinely useful while the financial returns disappoint investors.

Both things can be true.

What This Means for My Watchlist

Today’s news doesn’t make me want to chase every AI stock.

Quite the opposite.

It makes me more interested in understanding which companies have genuine advantages.

I’m looking at three questions:

First, who controls the infrastructure?

Computing capacity, energy availability and specialized hardware remain important competitive advantages.

Second, who can deliver AI at a sustainable cost?

Microsoft’s latest announcement reinforces the idea that efficiency could become a major differentiator.

Third, who gets paid when AI becomes essential?

Cybersecurity providers, enterprise software companies and infrastructure suppliers may have more predictable business models than some frontier-model developers.

That doesn’t mean they’re automatically good investments. Valuation still matters.

But I increasingly prefer businesses with identifiable customers, recurring demand and a credible path to cash flow.

My Take: The AI Industry Is Growing Up

If I had to summarize November 17 in one sentence, it would be this:

The AI race is shifting from proving what the technology can do to proving who can build a sustainable business around it.

Microsoft is competing on cost and independence.

Anthropic is preparing to face public-market scrutiny.

OpenAI is expanding into security-sensitive environments while facing questions about autonomous-agent safety.

And the infrastructure bill keeps growing.

I’m still convinced AI will change how we work, build software and operate businesses.

But I’m becoming increasingly skeptical of the idea that every company involved will be a winner.

The technology can succeed without every AI investment succeeding.

That’s probably the most important distinction to remember.

Tomorrow there will be another model, another benchmark and another billion-dollar announcement.

I’ll be watching the economics behind them.

Because that’s where the real story is starting to unfold.


Sources and Further Reading

These are my personal observations as a technology enthusiast and private investor. This article is for informational purposes only and does not constitute investment advice.

This is my personal opinion, not investment advice. I am not a licensed financial adviser. Do your own research. Disclosure: I may hold positions in assets mentioned here. Disclaimer

Get Thomas' briefing by email

One short email on days with a new article. Nothing on the other days.

More on AI