1. Introduction

Before coming to Canada to study IT, I spent many years working in management roles in China. My background is primarily in finance — I have been an accountant, a finance manager, and eventually a general manager responsible for operations, teams, and business strategy. Because of this experience, I tend to observe problems from a managerial and structural perspective rather than from surface-level impressions.

Over the past year, I have noticed a common sentiment shared by international students, domestic students, new graduates, and even mid-career professionals:
“Finding a job in tech has become much harder.”

This sentiment isn’t limited to one country or one school. It is global. North America, Europe, and Asia are all experiencing a similar shift. Internships are harder to get, junior roles are fewer, and companies appear more cautious than they were just two years ago. These are not speculative claims — they are observable facts reflected in job boards, offer rates, and conversations across the industry.

But acknowledging this phenomenon is not the purpose of this article.
My intention is much narrower:

To explore why this is happening — particularly from the perspective of business logic and organizational decision-making.

Before we move on, I want to clarify the scope:

The job market is shaped by multiple factors, including geopolitics, economic cycles, and technological disruptions.
I am not trying to quantify each factor’s exact weight.
I focus on AI because it is one area where I can offer a perspective shaped by management experience and firsthand usage.
Focusing on one factor does not imply it is the dominant one. It is simply the factor I can explain most clearly.

With that disclaimer, let’s break down the broader forces behind the current tech job landscape.

2. Multiple Factors Are Reshaping Today’s Tech Job Market

When people discuss the difficulty of finding CS and IT positions today, they often oversimplify the problem into one-sentence explanations:
“The economy is bad.”
“AI took the jobs.”
“Companies are moving work overseas.”
These statements are not wrong, but they are incomplete and lack nuance.

2.1 Global macroeconomic pressure

The world is undergoing a multi-dimensional set of disruptions:

Slower global economic growth
High interest rates across major economies
Reduced venture capital funding
De-risking and supply chain restructuring
Ongoing regional conflicts

These pressures naturally lead companies to reduce budgets, limit new hiring, and restructure departments.

2.2 Corporate strategy is shifting

Even large companies have become more cautious:

Big tech firms are optimizing teams
Startups are extending runway instead of expanding
Enterprises are prioritizing cost efficiency

These shifts affect hiring across all technical roles — from software engineering to IT support to cybersecurity.

2.3 AI as one factor among many

AI is part of this picture, but not the whole picture.

It is tempting to conclude “AI caused everything.”
But economic, political, and strategic forces existed long before AI began accelerating.

My discussion of AI is not an attempt to assign it the largest weight.
It is simply an attempt to articulate one important mechanism that I believe many people have misunderstood — especially how AI interacts with organizational decision-making.

This leads us to the next chapter.

3. Core Insight #1 — Understanding Why Companies Are Laying People Off

This is the part where my management background becomes relevant.

From the outside, layoffs look like a simple chain of events:

AI is getting better
Companies need fewer people
So companies cut staff

But internally, the logic is far more complex.

3.1 Companies are not laying people off because AI has fully matured

Rather:

Companies are laying people off because AI is partially usable.

This distinction is crucial.

If AI were fully mature — meaning it could replace entire roles, integrate seamlessly with systems, and perform tasks end-to-end — then companies would not be in a cautious “wait and observe” mode.
They would be aggressively restructuring entire departments.

But that is not what is happening today.

What companies are seeing now is something else:

AI can already automate some tasks
AI cannot yet take over full jobs
But the tasks it can automate are the “repeatable, low-risk, well-defined” parts of many roles

And that leads directly to the next point.

3.2 The 100-to-70 model: AI does not replace “jobs” — it replaces “task fragments”

In my YouTube video, I used a simple example:

Imagine a company with 100 employees.

After experimenting with AI tools, the leadership realizes:

AI can automate 20–40% of the repetitive or mechanical tasks performed by many employees.
These tasks are distributed across everyone, not isolated within a specific group.

So what happens is not: “AI replaces 30 people.” The real model looks like this:

Original state:

100 people perform 100% of the workload.

After AI partially intervenes:

70 people + AI ≈ 100 people’s original output.

Not because AI replaces 30 complete roles.
But because AI replaces 30% of the work spread across the entire department.

The tasks that disappear are not entire job descriptions — they are thousands of “small tasks”:
routine emails, formatting, documentation, data entry, code scaffolding, basic research, troubleshooting steps, etc.

When companies see this, the managerial logic becomes straightforward:

“We no longer need 100 people to produce the same output. We only need 70 people who can work with AI effectively.”

And because the pressure to cut costs is high, the company chooses the conservative option:

Reduce headcount now, explore full AI integration later.

3.3 Why companies cannot replace the remaining 70 people (yet)

Three reasons:

Reason #1: AI is powerful, but not standardized

Just like the early days of steam engines, the technology appears before the standards appear.

No universal workflow patterns
No unified integration interfaces
No stable expectations of output
No industry consensus on implementation

Without standardization, companies cannot build stable AI-driven workflows.

Reason #2: Management does not know how AI fits into existing processes

Leaders are asking questions like:

Should AI write documentation or only assist it?
Who is accountable when AI makes mistakes?
How should AI be integrated with existing ticketing systems, CRMs, ERPs, or codebases?
How much human review is required?

These questions have no simple answers today.

So companies take a “safe” approach:

Use AI for partial automation → reduce headcount → wait for the industry to figure out the rest.

Reason #3: New standards are emerging but not yet adopted

Technologies like MCP (Model Context Protocol) signal what the future may look like:

AI accessing local files
AI interacting with UI elements
AI controlling system workflows
AI operating as part of the OS or app ecosystem

But these capabilities are still early.
They require testing, governance, compliance, and industry alignment.

3.4 This is a transitional period — the “pre-standardization gap”

History repeats itself:

Capability appears
Adoption is inconsistent
Standards emerge
Entire industries reorganize

Right now, tech is in the chaotic middle stage:

AI is strong enough to reduce certain labor needs
AI is not mature enough to restructure workflows end-to-end
So companies shrink teams but do not fully rebuild them

From the perspective of job seekers, this is the worst stage to experience.

It is not the end of the tech industry.
It is the uncomfortable gap before the next transformation.

4. Core Insight #2 — The Widening, Exponential Gap in AI Usage Ability

Now we shift to the individual level.

Even if AI is only partially usable today, one phenomenon is absolutely clear:

The ability to use AI varies dramatically from person to person, and the gap is widening exponentially.

4.1 My own AI usage: likely within the top 5% (outside AI professionals)

I am not an AI researcher, not a software engineer, and not someone working in AI labs.
But compared to ordinary users, my AI usage pattern is extremely intensive:

I use tools like Sider to run parallel model comparisons
I use AI every day for learning, planning, architecture design, writing, and debugging
I ask 2,000–3,000 questions per month
My typical prompt includes context, reasoning steps, and expected output
The longest prompts I have written exceed 800–900 words
I integrate AI into everything:
network concepts, system design, homelab planning, career strategy, and academic learning

I am not special — I simply built strong habits of asking questions, testing ideas, and pushing models to their boundaries.

But this alone places me ahead of the majority of users.

4.2 The majority of people still use AI casually

Most people:

Ask AI a few questions per week
Use it like a fancier search engine
Do not build workflows
Do not combine multiple models
Do not perform iterative reasoning
Do not externalize their thinking
Do not test model limits or compare versions

This creates a structural divide:

A small group of people are amplifying themselves through AI
while a large group is barely benefiting from it.

4.3 This gap will widen year by year

Because:

The earlier you adopt workflow-level AI usage, the more compounding benefits you accumulate
AI literacy improves exponentially (not linearly)
Output quality improves as you refine your prompting and reasoning
The skill becomes a permanent differentiator in the job market

When companies must choose whom to retain:

They choose the people who can produce the output of 2–3 ordinary employees — the people who use AI not as a novelty, but as an operational tool.

This is why AI usage differences matter so much today.

5. What Individuals Should Do in This Transitional Era

Given this landscape, what can individuals — especially students and early-career professionals — do?

Here are my own conclusions, based on both industry observations and personal experience.

5.1 Stay aware of technological evolution (especially standard-level changes)

Learn about:

MCP
Workflow automation
Local AI
Model toolchains
Prompting patterns
AI-assisted development The people who understand these shifts early will have a major advantage when the next hiring wave arrives.

5.2 Strengthen your logical reasoning and learning abilities

AI can help you learn facts, but it cannot replace:

breaking down problems
asking the right questions
understanding dependencies
constructing workflows
evaluating trade-offs
making strategic decisions

These abilities will never lose relevance.

5.3 Actively integrate AI into your own processes

Don’t wait for companies to teach you how to use AI.
Begin building your own experiments:

Let AI optimize your study plan
Let AI help you build system diagrams
Let AI review your scripts or configs
Let AI help you document your homelab
Let AI support your content creation
Let AI analyze your mistakes and help you iterate

The key is not knowing AI exists —
the key is giving AI a place inside your daily workflow.

5.4 Be patient — the next cycle will come after standardization

Today’s job market is difficult because we are in the “messy middle”:

AI is emerging
But workflows aren’t built
Standards are forming
Companies are waiting

Once the standards solidify, the industry will need:

new roles
new tools
new infrastructure
new process designers
new workflows
new human–AI interfaces

The people who survive this transitional period —
who persist in learning and building —
will be the first to ride the next wave.

Conclusion

The tech job market feels difficult today not because the industry is dying,
but because it is transforming.

We are standing in the gap:

after capability appears
before standards stabilize
as companies reduce costs
while individuals adapt to AI
and before the next expansion cycle begins

My goal in writing this article is not to offer temporary comfort,
but to explain — from a managerial and structural perspective —
why this moment feels the way it does.

And more importantly:

Why the people who prepare now
will benefit the most when the transformation is complete.