October 28, 2025 • 31 mins
The source provides an extensive overview of Amazon’s significant corporate layoffs, announced in October 2025, which saw the elimination of approximately 14,000 white-collar jobs. These reductions, targeting departments such as human resources, devices, and Prime Video, are explicitly framed as part of a strategic pivot toward artificial intelligence (AI), aiming to reduce bureaucracy and achieve massive efficiency gains. The episode details how Amazon's post-pandemic overhiring, combined with CEO Andy Jassy's vision for AI-driven automation, led to the cuts, with internal documents suggesting AI could replace hundreds of thousands of roles in the coming years. Furthermore, the source examines the broader implications, including the types of jobs at risk across the tech sector and the contentious debate over whether AI will ultimately create more opportunities than it destroys.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Welcome to the deep dive. Today. We're tackling something huge,
a story that really feels like a turning point, maybe
less like a standard layoff and more like a signpost
for a whole new economic reality. We're diving deep into
Amazon's announcement from October twenty twenty five, fourteen thousand corporate
jobs eliminated, right, And.
Speaker 2 (00:22):
The significance here it's not just the number, though that's
big enough, it's the reason behind it.
Speaker 1 (00:26):
Exactly this time. It's not primarily about a recession or
a temporary downturn. The driver is artificial intelligence AI.
Speaker 2 (00:34):
That's absolutely right. And our mission in this deep dive
for you listening is really to go beyond those headlines.
We need to scrutinize the actual source material, the internal memos,
the financial reports, the projections, into the details, yet to
extract the specific facts, the nuggets of knowledge that show
how AI is now strategically restructuring the white collar workforce.
This isn't just one company. It feels like a precedent.
Speaker 1 (00:56):
Okay, So let's start with those immediate cold facts. October
twenty eighth, twenty twenty five. What exactly happen?
Speaker 2 (01:01):
Amazon announces plans to cut approximately fourteen thousand corporate jobs.
And it wasn't a phased approach thing. It began immediately,
pink slips delivered via email that very day.
Speaker 1 (01:13):
Wow, just like that. Brutal. And how does this compare
to pass cuts at Amazon?
Speaker 2 (01:19):
Well, it's the single biggest round since that really tough
period back for twenty twenty two to twenty twenty three.
You remember they shed about twenty seven thousand positions total
back then during that whole right sizing.
Speaker 1 (01:29):
Push, right, I remember that. So this fourteen thousand is
a major event for them, even after those earlier cuts.
Speaker 2 (01:35):
It is, and we need to get the scale right now.
Fourteen thousand. You might hear that and think, well, Amazon
employees over one point five million people worldwide seems small
percentage wise.
Speaker 1 (01:44):
Yeah, that's often the counter argument, But we have.
Speaker 2 (01:45):
To focus on the corporate structure here, the people in offices, engineers, HR,
marketing managers. That group is roughly three hundred and fifty
thousand strong.
Speaker 1 (01:54):
Okay, So against that number.
Speaker 2 (01:55):
Against that number, fourteen thousand is about four percent. Now
four percent might not sound earth shattering either, but what
it signals is a massive strategic, almost cultural shift, a
pivot towards radical efficiency powered by technology, getting rid of bureaucracy.
Speaker 1 (02:08):
And that leads us right into the central tension we
really need to explore today, doesn't it. These cuts are
happening precisely when Amazon is sinking well billions into AI
development and infrastructure huge sums. Yeah, So how do we
square that? Is AI this engine of growth that just
needs some temporary streamlining to get going, or is it
(02:31):
fundamentally a tool to permanently trim the workforce changing work itself.
Speaker 2 (02:35):
That's the core dilemma. Let's unpack the journey, the road
that led Amazon to this point right.
Speaker 1 (02:39):
To really get why twenty twenty five became this moment
of reckoning. We need to look back. Amazon's workforce journey
post pandemic was well, intents doesn't cover it. They didn't
just grow, They exploded.
Speaker 2 (02:50):
Oh absolutely. During the peak of the COVID nineteen pandemic.
Their global workforce just ballooned. Went from around eight hundred
thousand and early twenty twenty something like that, yeah, over
one point six million by twenty twenty one doubled, and
that was driven by just insane e commerce demand. They
had to scale up everything, warehouses, delivery, and critically the
corporate teams needed to manage all that frantic expansion.
Speaker 1 (03:11):
But that pace could last could as things started.
Speaker 2 (03:15):
Normalizing, No, it was unsustainable. As physical stores reopened and
shopping patterns kind of reverted, Amazon found itself by late
twenty twenty two just massively overstacked pretty much everywhere.
Speaker 1 (03:26):
And that's when the first big cut started, the right.
Speaker 2 (03:28):
Sizing exactly that kicked off the first major adjustment phase. Yeah,
November twenty twenty two, we saw the first ten thousand
jobs go. Then soon after early twenty twenty three, another
much larger round eighteen thousand positions cut.
Speaker 1 (03:43):
And where those initial cuts hit hardest.
Speaker 2 (03:45):
They mostly targeted areas that were maybe seen as more
experimental or had grown too fast, like the device's division
think Alexa hardware, parts of the retail organization, and also
significantly administrative roles in human resources.
Speaker 1 (03:58):
Okay, so this drive for efficient wasn't just a twenty
twenty five thing. The sources show was already underway for years, right,
like a slow burn before the AI firestorm.
Speaker 2 (04:07):
A sustained multi year project is exactly right. It wasn't
a one off purge throughout twenty twenty four, and leading
right into twenty twenty five, there were smaller, more targeted
cuts continuing hundreds here, hundreds there, whichjerrys were those we
saw cuts in prime Video, some from the integration of
MGM Studios the audio division, and even earlier in twenty
twenty five, before this big October announcement, they had already
(04:29):
trimmed about fifteen percent from specific teams with n HR, so.
Speaker 1 (04:32):
The pressure was already building. Amazon was clearly in this
intense efficiency mode before AI became the headline reason.
Speaker 2 (04:39):
Definitely, the AI factor seems to have acted like an
accelerant though. It took this ongoing efficiency project and put
it into hyperdrive.
Speaker 1 (04:46):
And this all connects back to the top, doesn't it
to the CEO Andy Jasse?
Speaker 2 (04:50):
Absolutely, You really have to link these cuts to Jasse's
broader agenda he took over from Jeff Bezos in twenty
twenty one, inheriting this incredibly complex, often layered, maybe even
bloated organization.
Speaker 1 (05:03):
One build for speed and scale above all else.
Speaker 2 (05:04):
For decades, right, and Jase's mandate, pretty clearly from the
start was to streamline it, to root out bureaucracy and
eliminate processes that weren't adding value. This has been his crusade.
Speaker 1 (05:14):
Did he use any specific tactics to push this before
the big AI focus? How did he hunt down that bureaucracy?
Speaker 2 (05:21):
He got quite specific. Actually, one really interesting example the
sources mentioned was the launch of the Bureaucracy Mailbox in
September twenty twenty four.
Speaker 1 (05:29):
A mailbox seriously.
Speaker 2 (05:32):
Yeah, basically an internal suggestion box, but focused entirely on inefficiency.
He invited employees to complain about wasteful meetings, slow processes,
unnecessary approvals.
Speaker 1 (05:43):
Did it work? Did people actually use it?
Speaker 2 (05:45):
Apparently? Yes? It generated over fifteen hundred complaints according to
the reports, and it wasn't just for show. It directly
led to about four hundred and fifty documented process changes
across the company.
Speaker 1 (05:56):
Wow. Okay, so tangible results.
Speaker 2 (05:58):
And there was more direct pressure too. He issued a
specific mandate boost employee to manager ratio is by fifteen
percent by the first quarter of twenty.
Speaker 1 (06:06):
Twenty five, meaning fewer managers.
Speaker 2 (06:08):
Fewer managers each supervising more people. The goal was explicitly
to flatten the hierarchy, remove those middle layers.
Speaker 1 (06:15):
That sounds intense. I also remember Amazon bringing in that
really strict five day return to office policy earlier in
twenty twenty five. Was that part of this headcount strategy
like a way to get people to quit.
Speaker 2 (06:26):
The sources strongly suggest it was. It seems to have
been a strategic attempt to drive headcount reduction through well
what HR calls natural attrition.
Speaker 1 (06:35):
Hoping people would just leave rather than come back.
Speaker 2 (06:37):
Full time exactly. The expectation was that employees who really
didn't want to be in the office five days a
week would vote with their feet that would help trim
the numbers without needing massive layoffs.
Speaker 1 (06:48):
But it didn't work or not enough.
Speaker 2 (06:51):
Apparently, it didn't achieve the target reduction figures they needed,
which ultimately forced their hand and paved the way for
the involuntary cuts the fourteen thousand we saw on October.
They basically ran out of other options to hit their
headcount goals.
Speaker 1 (07:04):
Okay, so attrition fails, the pressure still on, and then
comes the explicit AI pivot. This is where generative AI
really enters the chat, so to speak.
Speaker 2 (07:14):
It becomes the main narrative. Yes, now, Amazon's been an
AI pioneer for ages right their recommendation engine AWS machine
learning tools. That's old news.
Speaker 1 (07:22):
Sarah foundational stuff.
Speaker 2 (07:23):
But the generative AI wave really catalyzed by chat GPT's
launch in late twenty twenty two. That seemed to supercharge
their internal strategy. It gave them a new powerful tool
for that efficiency drive, and they move fast on it,
very aggressively. In twenty twenty three, they launched Amazon dead Rock.
That's their big platform for businesses wanting to build their
(07:44):
own custom generative AI models using AWS right enabling others.
And they didn't just build, they bought in too, pouring
huge money up to four billion dollars committed into Anthropic.
Speaker 1 (07:56):
The company behind Claude, the rival AI.
Speaker 2 (07:58):
Model exactly so, launching their own platform and investing heavily
in a leading competitor that shows a massive, multi pronged commitment.
It signals that while Jass always wanted efficiency, generative AI
suddenly offered a way to achieve it at a speed
and scale that just wasn't possible before.
Speaker 1 (08:14):
Okay, let's talk about the money now, because this AI
strategy isn't cheap. The scale of Amazon's financial commitment here
is well, it's staggering.
Speaker 2 (08:23):
It truly is, and we need to be clear about
what we're talking about financially, Amazon committed one hundred billion
dollars in capital expenditures. That's CAPEX and twenty twenty five alone.
Speaker 1 (08:31):
Right capex. It may be quickly explained for everyone listening
what that signifies, because it's not just spending money exactly.
Speaker 2 (08:37):
Capex. Capital expenditure isn't your day to day operating costs
like salaries or marketing. This is investment in long term
physical assets, buying equipment, machinery, building things, big tangible stuff,
big tangible stuff, and in Amazon's case, overwhelmingly, it means
constructing these enormous specialized data centers. Now one hundred billion
(08:59):
dollars in one year, that's just a colossal sum. It
dwarfs the GDP of many countries.
Speaker 1 (09:04):
It signals a massive long term bet. And it was
up from the year before.
Speaker 2 (09:07):
Too, significantly up from eighty three billion dollars in capex
in twenty twenty four, so a big jump.
Speaker 1 (09:12):
And where is this one hundred billion dollars going.
Speaker 2 (09:14):
Yeah, Physically, the sources are pretty specific. The vast majority
is earmarked for building these cutting edge data centers, mostly
in the US states like Mississippi, Indiana, Ohio, North Carolina
are mentioned.
Speaker 1 (09:25):
And these aren't small projects, no way.
Speaker 2 (09:28):
Each individual project is reportedly costing around ten billion dollars.
This spending spree underlines Amazon's ambition they want to be
the dominant cloud infrastructure provider for the entire AI revolution.
They're trying to out muscle Google, Microsoft and even Open Ai.
Speaker 1 (09:45):
So once they committed that kind of cash, CEO jasse
sends out this memo June twenty twenty five. And this
memo seems to have really laid out the vision connecting
the spending to the workforce.
Speaker 2 (09:56):
It did, it crystallized the strategy, and the language he
used was incredibly direct, almost historic in tone. He declared, quote,
this generation of AI is the most transformative technology we've
seen since the Internet.
Speaker 1 (10:08):
Wow, that's a big claim since the Internet.
Speaker 2 (10:10):
Huge claim. And he didn't stop there. He painted a
picture of the future workplace revolving around AI agents.
Speaker 1 (10:15):
AI agents like software.
Speaker 2 (10:16):
Bots essentially, yes, autonomous software programs designed to handle routine
cognitive tasks, things like forecasting inventory needs, managing customer service
chats online, automatically optimizing product pages on the website for
better sales.
Speaker 1 (10:29):
These are all things people do now. Analysts, planners, customer
service reps, managers.
Speaker 2 (10:34):
Precisely, these are existing white collar jobs, and Jasse was
explicit about the impact.
Speaker 1 (10:41):
You didn't try to sugarcoat it.
Speaker 2 (10:42):
Not. According to the reports, he apparently warned directly that
this AI shift would lead to quote fewer people doing
some of the jobs that are being done today, and
he stated it would quote reduce our total corporate workforce
over the next few years.
Speaker 1 (10:56):
When the CEO says that so plainly, yeah, it's not
speculation anymore, it's policy.
Speaker 2 (11:02):
It's confirmation of a deliberate strategic headcount reduction plan enabled
by technology. And digging deeper internal documents reviewed by The
New York Times around October twenty twenty five, they revealed
some really specific and frankly alarming automation targets.
Speaker 1 (11:17):
Okay, what did those internal projection show.
Speaker 2 (11:19):
Amazon's own automation team projected they could avoid hiring one
hundred and sixty thousand US workers by the year twenty
twenty seven.
Speaker 1 (11:26):
Avoid hiring that's different from laying off. But it's still
a huge number of potential jobs just not materializing exactly.
Speaker 2 (11:33):
And this wasn't just about headcount. It was driven by cost,
how by expanding the use of warehouse robots and deploying
more advanced AI in those corporate roles we just discussed.
Speaker 1 (11:42):
Was there a specific financial target tied to that one
hundred and sixty thousand figure.
Speaker 2 (11:47):
Yes, and this is maybe the most telling detail. The
goal tied to that projection was saving thirty cents per
item shipped.
Speaker 1 (11:54):
Thirty cents. That sounds tiny.
Speaker 2 (11:56):
It sounds tiny until you remember Amazon scale. They ship
billions and billions of items globally every year. That thirty
cents multiplied across that volume, it translates into potentially tens
of billions of dollars in annual operational savings.
Speaker 1 (12:10):
Okay, that's the motivation. Tens of billions.
Speaker 2 (12:13):
That's the cold hard driver behind this automation push, pure
massive cost savings.
Speaker 1 (12:18):
And the sources mentioned an even bigger, longer term goal
beyond twenty twenty seven.
Speaker 2 (12:24):
Yeah, there were broader, leaked plans indicating a much more
ambitious target, aiming to replace a staggering six hundred thousand
rolls with automation by the year twenty thirty Three's.
Speaker 1 (12:33):
This hundred thousand, that's what percentage of their workforce.
Speaker 2 (12:36):
That would represent about one third of their entire workforce
at the time. The plans were reportedly drafted.
Speaker 1 (12:41):
A third of the company automated within a decade. That's
not streamlining, that's a total transformation.
Speaker 2 (12:45):
It's a fundamental reshaping of the company, aiming for a
future where a huge chunk of both cognitive and physical
labor is done by machines and software. And connecting back
to the October cuts Beth Galletti's internal memo on the.
Speaker 1 (12:59):
Day she's the head of hr right.
Speaker 2 (13:01):
Head of People, Experience and Technology. Yes, her memo framed
the immediate fourteen thousand cuts very clearly designed to quote,
reduce bureaucracy, remove layers, and shift resources directly towards these
AI priorities. The goal to let Amazon quote innovate much faster.
Speaker 1 (13:18):
Okay, So this deliberate shift towards AI driven automation naturally
makes anyone listening, especially if they work in a corporate role,
ask is my job next? Let's break down which specific
departments got hit in October twenty twenty five, because there's
a pattern here, isn't there a profile of an AI
exposed job seems to be emerging.
Speaker 2 (13:36):
There's a very clear pattern, yes, highly predictive. Let's start
with human resources or as Amazon called it the People,
Experience and Technology division.
Speaker 1 (13:44):
They were hit hard, and that was on top of.
Speaker 2 (13:45):
Cuts they already took earlier in the year, exactly building
on that fifteen percent reduction from earlier in twenty twenty five,
the rationale seems purely operational. AI can now automate a
lot of the administrative, time consuming work that traditionally falls
to mid level HR managers and specialists.
Speaker 1 (14:03):
Like what kind of HR tasks specifically? I think a
lot of people assume HR needs that human touch.
Speaker 2 (14:08):
It does for complex issues, sure, but think about the
routine parts. AI can screen thousands of resumes almost instantly,
matching keywords and predicting candidate success based on data, far
faster than a human recruiter. That reduces the need for screeners.
Speaker 1 (14:23):
Okay, resume screening makes sense. What else?
Speaker 2 (14:25):
Performance reviews? Compliance tracking? AI agents can monitor communication patterns,
track project milestones against goals, check for compliance policy adherence continuously.
They can flag issues and generate reports automatically.
Speaker 1 (14:38):
So it handles the data gathering and basic analysis part right.
Speaker 2 (14:41):
It takes over much of the administrative burden the paperwork
side of HR, freeing up or eliminated the need for
humans who used to do that processing?
Speaker 1 (14:49):
Okay, department too, devices and services, that's Alexa Echo. Those
kinds of products. Cuts there hit software engineers and product managers.
Now that feels a bit weird. Shouldn't They need more
engineers for AI products.
Speaker 2 (15:04):
They are hiring some types of AI engineers, the really
high level ones. But these cuts targeted roles involved in more,
let's say, iterative updates and day to day product management.
Why because the way these AI products are built is changing.
Modern AI assistants like Alexa are designed to be self
improving through machine learning. They ingest usage data, analyze their
(15:25):
own performance, identify errors, and can even push out small
updates autonomously.
Speaker 1 (15:30):
Ah So if the product manages itself to some.
Speaker 2 (15:32):
Extent, you need fewer human product managers overseeing those minute
to minute iterative development cycles. The AI handles more the
refinement loop.
Speaker 1 (15:40):
Itself, got it next up operations, the core logistics, the
supply chain stuff. Where is AI making cuts there?
Speaker 2 (15:46):
It's targeting the human element of prediction and error correction.
The sources talk about predictive AI forecasting demand with ene
quote eer ee accuracy.
Speaker 1 (15:54):
Eerie accuracy. That sounds slightly.
Speaker 2 (15:57):
Unnerving, doesn't it think back to those mass of inventory
problems Amazon had post pandemic huge overstalking issues. AI minimizes
that risk. It crunches insane amounts of data past sales,
consumer trends, what or forecasts global shipping delays to predict
what customers will want, where and when with much greater precision.
Speaker 1 (16:18):
So lef need for human analysts trying to guess those
things exactly.
Speaker 2 (16:22):
It reduces the need for the teams of human planners
and logistics analysts whose entire job was basically managing uncertainty
and trying to prevent those costly inventory errors.
Speaker 1 (16:31):
And we see a similar logic in advertising and prime video.
Speaker 2 (16:34):
Absolutely same principle. AD optimization used to involve team as
people constantly tweaking campaigns, running ab tests, analyzing results. Now
AI tools can automate that entire feedback loop. They adjust adbids, targeting,
and creative elements in real time based on performance data,
needing far less human oversight.
Speaker 1 (16:51):
And for prime video, it's the recommendation engine.
Speaker 2 (16:54):
Yeah, those what to watch next suggestions, they're constantly being
refined by self learning algorithms. The AI optimizes itself, reducing
the need for large teams of human analysts who used
to manually fine tune those content recommendations.
Speaker 1 (17:09):
Okay, and finally, even Amazon Web Services AWS, the powerhouse,
the profit engine, the foundation for all this AI stuff.
Even they weren't totally immune.
Speaker 2 (17:21):
Even AWS sawcuts specifically in routine cloud support roles. Now,
to be clear, AWS is aggressively hiring top tier AI specialists,
data scientists, engineers building the next generation of services.
Speaker 1 (17:33):
But cutting the basic support right.
Speaker 2 (17:35):
Those frontline roles handling common technical queries from customers. Increasingly
automated advanced chatbots and AI agents can handle a growing
percentage of those routine support tickets, diagnose common problems, and
guide users through fixes.
Speaker 1 (17:48):
So trim the fat on the basic stuff to free
up budget for the high end AI talent.
Speaker 2 (17:52):
That seems to be the strategy across the board. Automate
the routine to invest in the complex.
Speaker 1 (17:57):
So, looking across these areas HR admin, device consideration, demand forecasting,
AD tuning, basic tech support, what's the common thread? What
really defines an AI exposed job? Right now for you listening.
Speaker 2 (18:10):
The common denominator is routine cognitive labor. It's work that
involves processing information, following rules, identifying patterns, and making predictions
based on data, but in a way that's largely repetitive
or standardized.
Speaker 1 (18:22):
And experts agree on this.
Speaker 2 (18:23):
Yes, Analysis from people like Carl Fred Oxford University confirms this.
These are the classic roles vulnerable to AI. Clerical tasks,
customer service interactions, basic data entry and analysis, even entry
level coding that involves generating fairly standard blocks of code.
Speaker 1 (18:39):
You mentioned large language models lms. How are they different
from the algorithms we've had for years? Why is this
wave of AI so potent against these jobs?
Speaker 2 (18:48):
It's the generative aspect. Older algorithms mostly followed pre defined rules.
If this, then do that. Llms can understand natural language,
synthesize information from fast data sets, and generate new con
to emails, reports, codes, summaries, conversational responses that seems human like,
they'll just fall with script exactly. They can handle nuance,
manage multi step tasks, access and summarize information from different sources.
(19:12):
That's why research from places like Stanford shows that hiring
for entry level cognitive roles dropped what was it thirteen percent?
Speaker 1 (19:19):
Yeah, thirteen percent since lms became widespread.
Speaker 2 (19:22):
Because the llms can now do a chunk of that
entry level work.
Speaker 1 (19:25):
So Jasse's push for scrappier teams, it sounds like AI
is the muscle making that happen.
Speaker 2 (19:32):
It's the technological enabler. Absolutely flattening Hierarchyes, becomes much easier
when AI can take over some of the oversight and
reporting functions previously done by junior or mid level manager.
So well, if AI can track per progress, summarize team communications,
flag potential issues, and even delegate routine tasks automatically, then
one human manager can effectively oversee a larger team. The
(19:55):
AI handles the low level administrative oversight, freeing the human
manager for more strategic thinking, or simply allowing the company
to have fewer managers overall.
Speaker 1 (20:04):
Okay, but let's balance this. Amidst these fourteen thousand cuts,
where is Amazon spending money and actively hiring people. They're
not just cutting.
Speaker 2 (20:12):
Yeah, no, they're definitely hiring, but very strategically. They're focusing
on what they call key strategic areas. These are roles
that require deep specialized knowledge, complex problem solving, creativity, things
currently well beyond AI's capability. Advanced AI research and engineering,
sophisticated data science, specialized robotics development, and maintenance. Think about
(20:34):
that big UK investment you hear about forty billion over
three years that promises thousands of jobs. Yes, but they're
largely in building out fulfillment centers and the high tech
AI infrastructure itself.
Speaker 1 (20:45):
So jobs are being created, but they're different kinds of jobs.
Speaker 2 (20:48):
Very different. And the sources emphasize the overall outcome is
still aimed at creating a leaner, meaner Amazon, highly optimized,
highly automated. Even with new hiring in some areas, the
total the whole number of corporate workers relative to revenue
is expected to decrease significantly over time. And it's crucial
to understand this isn't just an Amazon story. What they're
(21:08):
doing is basically writing the playbook for the rest of
the industry, maybe the whole economy. We need to look
at the immediate ripples, the fallout, and the feeling inside
Amazon when this happened.
Speaker 1 (21:18):
Yeah, the sources described a really tense atmosphere, people walking
on pins and needles, waiting for that email to land,
and getting the news via personal email, sometimes at dawn.
That sounds incredibly impersonal and stressful.
Speaker 2 (21:32):
Deeply unsettling, and it creates this profound cognitive dissonance, especially
for the tech workers themselves. There was that quote, I
think from an HR specialist speaking anonymously. Oh yeah, the
paradox of the builder exactly, We're building the tools that
replace us. Imagine spending your career perfecting a technology only
to see it turned around and use to eliminate the
very function you perform. That's a fundamental challenge to the
(21:55):
old social contract between employer and employee.
Speaker 1 (21:58):
And this isn't happening at a vacuum. The tech sector was
already seeing significant layoffs in twenty twenty five even before
this Amazon news hit.
Speaker 2 (22:05):
Oh yeah, the trend was already accelerating layoffs. DOTIFYI, which
tracks this stuff, reported over ninety eight thousand tech job
cuts globally in twenty twenty five prior to Amazon's October announcement.
Amazon is often the bellweather, but they're definitely not alone.
Speaker 1 (22:20):
Who else was making similar moves explicitly citing AI.
Speaker 2 (22:23):
Well. Microsoft shed fifteen thousand rolls in the summer of
twenty twenty five, salesforce cut four thousand in September, and
their executives made that really striking.
Speaker 1 (22:31):
Claim, the one about AI doing half their work.
Speaker 2 (22:34):
Yeah, claiming AI now handles fifty percent of their operational tasks.
Think about what that implies for sales, marketing, customer support roles.
Speaker 1 (22:40):
What does that mean practically? If AI is doing fifty
percent of the work at salesforce, it means.
Speaker 2 (22:45):
The routine stuff drafting initial emails, researching leads, analyzing customal data,
summarizing calls, maybe even generating first draft proposals is increasingly automated.
The human salesperson shifts towards higher level tasks relationship building,
complex negotiation, closing deals, strategic account management.
Speaker 1 (23:04):
Less admin, more strategy, but needing fewer people overall to
hit the same target.
Speaker 2 (23:09):
That's the inevitable consequence. And it's not just software companies.
Look at ups acts fourteen thousand managers partly attributed to
AI optimizing routes and logistics. Luftanza cut four thousand jobs,
citing AI efficiencies and planning and scheduling. It's spreading.
Speaker 1 (23:23):
So there's this interesting economic picture. The sources mentioned that
in September twenty twenty five, the overall US job market
looked okay, adding two hundred and fifty four thousand jobs,
but unemployments still ticked up slightly to four point three percent.
How does that work?
Speaker 2 (23:35):
Yeah, Axios pointed this out. They called it the AI
labor market squeeze. Even if the broader economy is adding
jobs maybe in services, construction, healthcare, the big white collar employers,
the tech giants like Amazon, Microsoft, Google, they're realizing they
don't need to hire as aggressively for those traditional entry
level or mid level corporate roles.
Speaker 1 (23:56):
Anymore because AI is filling the productivity gap or.
Speaker 2 (24:00):
Handling the tasks those roles used to do. So you
get this weird situation. Overall job growth, but a tightening
or even shrinking in the specific sectors that used to
absorb lots of college graduates into cognitive work. The traditional
pathways are narrowing.
Speaker 1 (24:14):
Which brings us squarely to the big debate job destruction
versus job creation. The World Economic Forums twenty twenty five
survey had some pretty dramatic numbers on this.
Speaker 2 (24:23):
The numbers are stark, definitely on the destruction side. The
projection was ninety two million jobs lost globally to AI
by twenty thirty.
Speaker 1 (24:30):
Ninety two million. And these are the kinds of jobs
we've been talking.
Speaker 2 (24:33):
About exactly, clerical work, administrative support, customer service. They projected
chatbots could handle over fifty percent of queries. Basic coding,
routine data analysis, even some layers of middle management focused
on oversight. Anything fundamentally based on repetitive information processing is
at risk.
Speaker 1 (24:51):
But there's always the flip side, right, the argument that
technology creates new jobs.
Speaker 2 (24:55):
That's the counter narrative, and the WEF data supports that too,
at least numerical. They predicted the creation of one hundred
and seventy million new jobs by twenty.
Speaker 1 (25:04):
Thirty, more created than destroyed.
Speaker 2 (25:06):
Then mathematically yes, but these new jobs are very different.
They're in fields like AI ethics, data curation essentially teaching
and refining the AI models, prompt engineering, designing human AI
collaboration workflows, AI system maintenance, highly specialized roles.
Speaker 1 (25:23):
So the net number might be positive, but the skills
don't match up easily.
Speaker 2 (25:26):
That's the massive challenge, the transition risk. It's not a
simple swap. The skills needed for those one hundred and
seventy million new jobs are often vastly different from the
skills possessed by the ninety two million people whose routine
jobs are disappearing, and.
Speaker 1 (25:38):
Who gets hit hardest during this transition. The Burning Glass
Institute had a warning about this being a long process.
Speaker 2 (25:45):
They warn't about a multi decade shakeup, and the data
suggests it's hitting younger workers those just starting out particularly hard.
Right now, Remember that thirteen percent drop and hiring for
entry level coders right that's because the tasks usually given
to junior employees are often the easiest ones to automate
with current AI tools. The challenge is that our current
(26:06):
education systems and corporate training programs are struggling to upskill
people fast enough to meet the demand for these new,
complex AI related roles. There's a growing mismatch.
Speaker 1 (26:16):
This feels like it has really serious social and ethical implications,
especially the difference between this wave of automation and past ones.
Speaker 2 (26:24):
It's a critical distinction. Historically, automation primarily hit blue collar
manual labor in factories, while disruptive society could argue those
jobs were often physically demanding or repetitive. AI is different
because it's targeting cognitive, white collar work, the.
Speaker 1 (26:39):
Jobs that build the modern middle class exactly.
Speaker 2 (26:42):
The analysts, the supervisor, the office worker, the entry level
professional roles. These provided pathways to stable careers, and ironically,
many low skill, physically demanding jobs like warehouse work seem
more resilient. For now, Amazon still needs two hundred and
fifty thousand seasonal hires for packing boxes during the holidays.
Speaker 1 (27:01):
So the corporate ladder is changing fundamentally. You need AI
fluency now to even get on it or stay.
Speaker 2 (27:07):
On it, it seems that way, which risks increasing inequality.
If the pathway to stable middle class corporate work now
requires advanced AI skills that many people lack access to
or can't quickly acquire, you could see a further stratification
of the workforce.
Speaker 1 (27:22):
That warning from five or CEO Mikah Kaufman comes to
mind that AI is coming for everyone. Programmer, lawyer, salesperson.
No white collar role based on routine information processing seems
entirely safe.
Speaker 2 (27:33):
That's the paradigm shift Amazon is helping to accelerate. High
pay or a fancy title doesn't offer protection if the
core function of your job involves tasks that an AI
can perform more efficiently or cheaply. And this forces huge
questions onto policymakers like what like do we need to
seriously consider things like universal basic income? What kind of
massive government supported retraining programs are needed? How do we
(27:56):
ensure the benefits of AI productivity gains are shared broadly
captured by capital owners. These policy debates are lagging way
behind the technological reality Amazon is creating.
Speaker 1 (28:07):
So given all this disruption, what is Amazon actually doing
for the fourteen thousand people they let go? In October
they talked about transition support? What does that practically mean?
Speaker 2 (28:16):
Well, they're offering a package. It includes a period I
think ninety days for affected employees to try and find
another role within Amazon if possible, plus severance pay, access
to outplacement services to help find external jobs, and extended
health benefits.
Speaker 1 (28:30):
Is there any data on how effective that support was
after the earlier twenty twenty two to twenty three cuts.
Speaker 2 (28:36):
The data they released suggests it was reasonably effective relatively speaking.
They reported that around seventy five percent of those laid
off back then either found new roles or left the
company with financial packages they accepted.
Speaker 1 (28:47):
Okay, and what about upskilling for the remaining workforce? Are
they investing there?
Speaker 2 (28:51):
Yes, they're making a significant investment internally. The figure mentioned
is two point five billion dollars specifically for upskilling initiatives.
The stated goal is ambition to help prepare fifty million
people globally, not just employees for working in the AI era,
by twenty thirty.
Speaker 1 (29:07):
Fifty million people. That's huge.
Speaker 2 (29:08):
It's a massive goal. And Jasse's advice to his own
remaining employees was very direct. Educate yourself, experiment with AI.
The message is clear, adapt or risk becoming obsolete. Hashtag
tax tag outro Okay.
Speaker 1 (29:22):
So let's try to summarize the math here. Amazon's fourteen
thousand cuts in October twenty twenty five weren't just random downsizing.
They were a direct, calculated consequence of a strategy to
optimize ruthlessly for AI efficiency and finally deliver on Andy
Jasse's long term goal of slashing bureaucracy. They're actively building
that nimbler.
Speaker 2 (29:41):
Giant exactly, and in doing so, that nimbler giant is
essentially validating the fear that's been lurking in the background
for years. It confirms that yes, AI is a powerful
tool for enabling large scale workforce reductions, all justified in
the name of efficiency and innovation and shareholder value.
Speaker 1 (29:55):
It sets a precedent.
Speaker 2 (29:56):
It absolutely sets a precedent for the entire white collar economy.
When a company like Amazon pouring one hundred billion dollars
into AI infrastructure simultaneously cuts thousands of corporate jobs explicitly
because of AI's potential. That sends a powerful signal. It
forces those crucial policy questions about worker safety nets, retraining mandates,
(30:16):
the future of work itself, questions that frankly, we don't
have good answers for yet.
Speaker 1 (30:20):
So the final takeaway for you listening today seems pretty stark.
This shift isn't coming, it's here. It demands urgent adaptation.
It means actively moving beyond easily automated routine tasks. It
means learning things like trompt engineering, figuring out how to
use AI tools as collaborators, not competitors, and really focusing
on developing those skills that remain uniquely human. Deep critical thinking,
(30:44):
genuine creativity, complex problem solving, and sophisticated emotional intelligence and empathy.
Speaker 2 (30:49):
Couldn't agree more. We know Amazon itself will be reinvented
that internal projection aiming to replace six hundred thousand rolls
by twenty thirty three. That seals Amazon's future trends information.
But the really critical question, the provocative thought may be
for you to take away builds on Jasse's own words.
He said those who embrace the change will be well positioned.
(31:09):
But given the sheer scale of automation planned the systemic
pressure this playbook puts on every company, what happens to
everyone who doesn't get retrained effectively or quickly enough as
this model becomes standard practice, Who inevitably gets left behind
in this AI driven restructuring of the economy. That's the
profound challenge we're all facing now
Welcome to the deep dive. Today. We're tackling something huge,
a story that really feels like a turning point, maybe
less like a standard layoff and more like a signpost
for a whole new economic reality. We're diving deep into
Amazon's announcement from October twenty twenty five, fourteen thousand corporate
jobs eliminated, right, And.
Speaker 2 (00:22):
The significance here it's not just the number, though that's
big enough, it's the reason behind it.
Speaker 1 (00:26):
Exactly this time. It's not primarily about a recession or
a temporary downturn. The driver is artificial intelligence AI.
Speaker 2 (00:34):
That's absolutely right. And our mission in this deep dive
for you listening is really to go beyond those headlines.
We need to scrutinize the actual source material, the internal memos,
the financial reports, the projections, into the details, yet to
extract the specific facts, the nuggets of knowledge that show
how AI is now strategically restructuring the white collar workforce.
This isn't just one company. It feels like a precedent.
Speaker 1 (00:56):
Okay, So let's start with those immediate cold facts. October
twenty eighth, twenty twenty five. What exactly happen?
Speaker 2 (01:01):
Amazon announces plans to cut approximately fourteen thousand corporate jobs.
And it wasn't a phased approach thing. It began immediately,
pink slips delivered via email that very day.
Speaker 1 (01:13):
Wow, just like that. Brutal. And how does this compare
to pass cuts at Amazon?
Speaker 2 (01:19):
Well, it's the single biggest round since that really tough
period back for twenty twenty two to twenty twenty three.
You remember they shed about twenty seven thousand positions total
back then during that whole right sizing.
Speaker 1 (01:29):
Push, right, I remember that. So this fourteen thousand is
a major event for them, even after those earlier cuts.
Speaker 2 (01:35):
It is, and we need to get the scale right now.
Fourteen thousand. You might hear that and think, well, Amazon
employees over one point five million people worldwide seems small
percentage wise.
Speaker 1 (01:44):
Yeah, that's often the counter argument, But we have.
Speaker 2 (01:45):
To focus on the corporate structure here, the people in offices, engineers, HR,
marketing managers. That group is roughly three hundred and fifty
thousand strong.
Speaker 1 (01:54):
Okay, So against that number.
Speaker 2 (01:55):
Against that number, fourteen thousand is about four percent. Now
four percent might not sound earth shattering either, but what
it signals is a massive strategic, almost cultural shift, a
pivot towards radical efficiency powered by technology, getting rid of bureaucracy.
Speaker 1 (02:08):
And that leads us right into the central tension we
really need to explore today, doesn't it. These cuts are
happening precisely when Amazon is sinking well billions into AI
development and infrastructure huge sums. Yeah, So how do we
square that? Is AI this engine of growth that just
needs some temporary streamlining to get going, or is it
(02:31):
fundamentally a tool to permanently trim the workforce changing work itself.
Speaker 2 (02:35):
That's the core dilemma. Let's unpack the journey, the road
that led Amazon to this point right.
Speaker 1 (02:39):
To really get why twenty twenty five became this moment
of reckoning. We need to look back. Amazon's workforce journey
post pandemic was well, intents doesn't cover it. They didn't
just grow, They exploded.
Speaker 2 (02:50):
Oh absolutely. During the peak of the COVID nineteen pandemic.
Their global workforce just ballooned. Went from around eight hundred
thousand and early twenty twenty something like that, yeah, over
one point six million by twenty twenty one doubled, and
that was driven by just insane e commerce demand. They
had to scale up everything, warehouses, delivery, and critically the
corporate teams needed to manage all that frantic expansion.
Speaker 1 (03:11):
But that pace could last could as things started.
Speaker 2 (03:15):
Normalizing, No, it was unsustainable. As physical stores reopened and
shopping patterns kind of reverted, Amazon found itself by late
twenty twenty two just massively overstacked pretty much everywhere.
Speaker 1 (03:26):
And that's when the first big cut started, the right.
Speaker 2 (03:28):
Sizing exactly that kicked off the first major adjustment phase. Yeah,
November twenty twenty two, we saw the first ten thousand
jobs go. Then soon after early twenty twenty three, another
much larger round eighteen thousand positions cut.
Speaker 1 (03:43):
And where those initial cuts hit hardest.
Speaker 2 (03:45):
They mostly targeted areas that were maybe seen as more
experimental or had grown too fast, like the device's division
think Alexa hardware, parts of the retail organization, and also
significantly administrative roles in human resources.
Speaker 1 (03:58):
Okay, so this drive for efficient wasn't just a twenty
twenty five thing. The sources show was already underway for years, right,
like a slow burn before the AI firestorm.
Speaker 2 (04:07):
A sustained multi year project is exactly right. It wasn't
a one off purge throughout twenty twenty four, and leading
right into twenty twenty five, there were smaller, more targeted
cuts continuing hundreds here, hundreds there, whichjerrys were those we
saw cuts in prime Video, some from the integration of
MGM Studios the audio division, and even earlier in twenty
twenty five, before this big October announcement, they had already
(04:29):
trimmed about fifteen percent from specific teams with n HR, so.
Speaker 1 (04:32):
The pressure was already building. Amazon was clearly in this
intense efficiency mode before AI became the headline reason.
Speaker 2 (04:39):
Definitely, the AI factor seems to have acted like an
accelerant though. It took this ongoing efficiency project and put
it into hyperdrive.
Speaker 1 (04:46):
And this all connects back to the top, doesn't it
to the CEO Andy Jasse?
Speaker 2 (04:50):
Absolutely, You really have to link these cuts to Jasse's
broader agenda he took over from Jeff Bezos in twenty
twenty one, inheriting this incredibly complex, often layered, maybe even
bloated organization.
Speaker 1 (05:03):
One build for speed and scale above all else.
Speaker 2 (05:04):
For decades, right, and Jase's mandate, pretty clearly from the
start was to streamline it, to root out bureaucracy and
eliminate processes that weren't adding value. This has been his crusade.
Speaker 1 (05:14):
Did he use any specific tactics to push this before
the big AI focus? How did he hunt down that bureaucracy?
Speaker 2 (05:21):
He got quite specific. Actually, one really interesting example the
sources mentioned was the launch of the Bureaucracy Mailbox in
September twenty twenty four.
Speaker 1 (05:29):
A mailbox seriously.
Speaker 2 (05:32):
Yeah, basically an internal suggestion box, but focused entirely on inefficiency.
He invited employees to complain about wasteful meetings, slow processes,
unnecessary approvals.
Speaker 1 (05:43):
Did it work? Did people actually use it?
Speaker 2 (05:45):
Apparently? Yes? It generated over fifteen hundred complaints according to
the reports, and it wasn't just for show. It directly
led to about four hundred and fifty documented process changes
across the company.
Speaker 1 (05:56):
Wow. Okay, so tangible results.
Speaker 2 (05:58):
And there was more direct pressure too. He issued a
specific mandate boost employee to manager ratio is by fifteen
percent by the first quarter of twenty.
Speaker 1 (06:06):
Twenty five, meaning fewer managers.
Speaker 2 (06:08):
Fewer managers each supervising more people. The goal was explicitly
to flatten the hierarchy, remove those middle layers.
Speaker 1 (06:15):
That sounds intense. I also remember Amazon bringing in that
really strict five day return to office policy earlier in
twenty twenty five. Was that part of this headcount strategy
like a way to get people to quit.
Speaker 2 (06:26):
The sources strongly suggest it was. It seems to have
been a strategic attempt to drive headcount reduction through well
what HR calls natural attrition.
Speaker 1 (06:35):
Hoping people would just leave rather than come back.
Speaker 2 (06:37):
Full time exactly. The expectation was that employees who really
didn't want to be in the office five days a
week would vote with their feet that would help trim
the numbers without needing massive layoffs.
Speaker 1 (06:48):
But it didn't work or not enough.
Speaker 2 (06:51):
Apparently, it didn't achieve the target reduction figures they needed,
which ultimately forced their hand and paved the way for
the involuntary cuts the fourteen thousand we saw on October.
They basically ran out of other options to hit their
headcount goals.
Speaker 1 (07:04):
Okay, so attrition fails, the pressure still on, and then
comes the explicit AI pivot. This is where generative AI
really enters the chat, so to speak.
Speaker 2 (07:14):
It becomes the main narrative. Yes, now, Amazon's been an
AI pioneer for ages right their recommendation engine AWS machine
learning tools. That's old news.
Speaker 1 (07:22):
Sarah foundational stuff.
Speaker 2 (07:23):
But the generative AI wave really catalyzed by chat GPT's
launch in late twenty twenty two. That seemed to supercharge
their internal strategy. It gave them a new powerful tool
for that efficiency drive, and they move fast on it,
very aggressively. In twenty twenty three, they launched Amazon dead Rock.
That's their big platform for businesses wanting to build their
(07:44):
own custom generative AI models using AWS right enabling others.
And they didn't just build, they bought in too, pouring
huge money up to four billion dollars committed into Anthropic.
Speaker 1 (07:56):
The company behind Claude, the rival AI.
Speaker 2 (07:58):
Model exactly so, launching their own platform and investing heavily
in a leading competitor that shows a massive, multi pronged commitment.
It signals that while Jass always wanted efficiency, generative AI
suddenly offered a way to achieve it at a speed
and scale that just wasn't possible before.
Speaker 1 (08:14):
Okay, let's talk about the money now, because this AI
strategy isn't cheap. The scale of Amazon's financial commitment here
is well, it's staggering.
Speaker 2 (08:23):
It truly is, and we need to be clear about
what we're talking about financially, Amazon committed one hundred billion
dollars in capital expenditures. That's CAPEX and twenty twenty five alone.
Speaker 1 (08:31):
Right capex. It may be quickly explained for everyone listening
what that signifies, because it's not just spending money exactly.
Speaker 2 (08:37):
Capex. Capital expenditure isn't your day to day operating costs
like salaries or marketing. This is investment in long term
physical assets, buying equipment, machinery, building things, big tangible stuff,
big tangible stuff, and in Amazon's case, overwhelmingly, it means
constructing these enormous specialized data centers. Now one hundred billion
(08:59):
dollars in one year, that's just a colossal sum. It
dwarfs the GDP of many countries.
Speaker 1 (09:04):
It signals a massive long term bet. And it was
up from the year before.
Speaker 2 (09:07):
Too, significantly up from eighty three billion dollars in capex
in twenty twenty four, so a big jump.
Speaker 1 (09:12):
And where is this one hundred billion dollars going.
Speaker 2 (09:14):
Yeah, Physically, the sources are pretty specific. The vast majority
is earmarked for building these cutting edge data centers, mostly
in the US states like Mississippi, Indiana, Ohio, North Carolina
are mentioned.
Speaker 1 (09:25):
And these aren't small projects, no way.
Speaker 2 (09:28):
Each individual project is reportedly costing around ten billion dollars.
This spending spree underlines Amazon's ambition they want to be
the dominant cloud infrastructure provider for the entire AI revolution.
They're trying to out muscle Google, Microsoft and even Open Ai.
Speaker 1 (09:45):
So once they committed that kind of cash, CEO jasse
sends out this memo June twenty twenty five. And this
memo seems to have really laid out the vision connecting
the spending to the workforce.
Speaker 2 (09:56):
It did, it crystallized the strategy, and the language he
used was incredibly direct, almost historic in tone. He declared, quote,
this generation of AI is the most transformative technology we've
seen since the Internet.
Speaker 1 (10:08):
Wow, that's a big claim since the Internet.
Speaker 2 (10:10):
Huge claim. And he didn't stop there. He painted a
picture of the future workplace revolving around AI agents.
Speaker 1 (10:15):
AI agents like software.
Speaker 2 (10:16):
Bots essentially, yes, autonomous software programs designed to handle routine
cognitive tasks, things like forecasting inventory needs, managing customer service
chats online, automatically optimizing product pages on the website for
better sales.
Speaker 1 (10:29):
These are all things people do now. Analysts, planners, customer
service reps, managers.
Speaker 2 (10:34):
Precisely, these are existing white collar jobs, and Jasse was
explicit about the impact.
Speaker 1 (10:41):
You didn't try to sugarcoat it.
Speaker 2 (10:42):
Not. According to the reports, he apparently warned directly that
this AI shift would lead to quote fewer people doing
some of the jobs that are being done today, and
he stated it would quote reduce our total corporate workforce
over the next few years.
Speaker 1 (10:56):
When the CEO says that so plainly, yeah, it's not
speculation anymore, it's policy.
Speaker 2 (11:02):
It's confirmation of a deliberate strategic headcount reduction plan enabled
by technology. And digging deeper internal documents reviewed by The
New York Times around October twenty twenty five, they revealed
some really specific and frankly alarming automation targets.
Speaker 1 (11:17):
Okay, what did those internal projection show.
Speaker 2 (11:19):
Amazon's own automation team projected they could avoid hiring one
hundred and sixty thousand US workers by the year twenty
twenty seven.
Speaker 1 (11:26):
Avoid hiring that's different from laying off. But it's still
a huge number of potential jobs just not materializing exactly.
Speaker 2 (11:33):
And this wasn't just about headcount. It was driven by cost,
how by expanding the use of warehouse robots and deploying
more advanced AI in those corporate roles we just discussed.
Speaker 1 (11:42):
Was there a specific financial target tied to that one
hundred and sixty thousand figure.
Speaker 2 (11:47):
Yes, and this is maybe the most telling detail. The
goal tied to that projection was saving thirty cents per
item shipped.
Speaker 1 (11:54):
Thirty cents. That sounds tiny.
Speaker 2 (11:56):
It sounds tiny until you remember Amazon scale. They ship
billions and billions of items globally every year. That thirty
cents multiplied across that volume, it translates into potentially tens
of billions of dollars in annual operational savings.
Speaker 1 (12:10):
Okay, that's the motivation. Tens of billions.
Speaker 2 (12:13):
That's the cold hard driver behind this automation push, pure
massive cost savings.
Speaker 1 (12:18):
And the sources mentioned an even bigger, longer term goal
beyond twenty twenty seven.
Speaker 2 (12:24):
Yeah, there were broader, leaked plans indicating a much more
ambitious target, aiming to replace a staggering six hundred thousand
rolls with automation by the year twenty thirty Three's.
Speaker 1 (12:33):
This hundred thousand, that's what percentage of their workforce.
Speaker 2 (12:36):
That would represent about one third of their entire workforce
at the time. The plans were reportedly drafted.
Speaker 1 (12:41):
A third of the company automated within a decade. That's
not streamlining, that's a total transformation.
Speaker 2 (12:45):
It's a fundamental reshaping of the company, aiming for a
future where a huge chunk of both cognitive and physical
labor is done by machines and software. And connecting back
to the October cuts Beth Galletti's internal memo on the.
Speaker 1 (12:59):
Day she's the head of hr right.
Speaker 2 (13:01):
Head of People, Experience and Technology. Yes, her memo framed
the immediate fourteen thousand cuts very clearly designed to quote,
reduce bureaucracy, remove layers, and shift resources directly towards these
AI priorities. The goal to let Amazon quote innovate much faster.
Speaker 1 (13:18):
Okay, So this deliberate shift towards AI driven automation naturally
makes anyone listening, especially if they work in a corporate role,
ask is my job next? Let's break down which specific
departments got hit in October twenty twenty five, because there's
a pattern here, isn't there a profile of an AI
exposed job seems to be emerging.
Speaker 2 (13:36):
There's a very clear pattern, yes, highly predictive. Let's start
with human resources or as Amazon called it the People,
Experience and Technology division.
Speaker 1 (13:44):
They were hit hard, and that was on top of.
Speaker 2 (13:45):
Cuts they already took earlier in the year, exactly building
on that fifteen percent reduction from earlier in twenty twenty five,
the rationale seems purely operational. AI can now automate a
lot of the administrative, time consuming work that traditionally falls
to mid level HR managers and specialists.
Speaker 1 (14:03):
Like what kind of HR tasks specifically? I think a
lot of people assume HR needs that human touch.
Speaker 2 (14:08):
It does for complex issues, sure, but think about the
routine parts. AI can screen thousands of resumes almost instantly,
matching keywords and predicting candidate success based on data, far
faster than a human recruiter. That reduces the need for screeners.
Speaker 1 (14:23):
Okay, resume screening makes sense. What else?
Speaker 2 (14:25):
Performance reviews? Compliance tracking? AI agents can monitor communication patterns,
track project milestones against goals, check for compliance policy adherence continuously.
They can flag issues and generate reports automatically.
Speaker 1 (14:38):
So it handles the data gathering and basic analysis part right.
Speaker 2 (14:41):
It takes over much of the administrative burden the paperwork
side of HR, freeing up or eliminated the need for
humans who used to do that processing?
Speaker 1 (14:49):
Okay, department too, devices and services, that's Alexa Echo. Those
kinds of products. Cuts there hit software engineers and product managers.
Now that feels a bit weird. Shouldn't They need more
engineers for AI products.
Speaker 2 (15:04):
They are hiring some types of AI engineers, the really
high level ones. But these cuts targeted roles involved in more,
let's say, iterative updates and day to day product management.
Why because the way these AI products are built is changing.
Modern AI assistants like Alexa are designed to be self
improving through machine learning. They ingest usage data, analyze their
(15:25):
own performance, identify errors, and can even push out small
updates autonomously.
Speaker 1 (15:30):
Ah So if the product manages itself to some.
Speaker 2 (15:32):
Extent, you need fewer human product managers overseeing those minute
to minute iterative development cycles. The AI handles more the
refinement loop.
Speaker 1 (15:40):
Itself, got it next up operations, the core logistics, the
supply chain stuff. Where is AI making cuts there?
Speaker 2 (15:46):
It's targeting the human element of prediction and error correction.
The sources talk about predictive AI forecasting demand with ene
quote eer ee accuracy.
Speaker 1 (15:54):
Eerie accuracy. That sounds slightly.
Speaker 2 (15:57):
Unnerving, doesn't it think back to those mass of inventory
problems Amazon had post pandemic huge overstalking issues. AI minimizes
that risk. It crunches insane amounts of data past sales,
consumer trends, what or forecasts global shipping delays to predict
what customers will want, where and when with much greater precision.
Speaker 1 (16:18):
So lef need for human analysts trying to guess those
things exactly.
Speaker 2 (16:22):
It reduces the need for the teams of human planners
and logistics analysts whose entire job was basically managing uncertainty
and trying to prevent those costly inventory errors.
Speaker 1 (16:31):
And we see a similar logic in advertising and prime video.
Speaker 2 (16:34):
Absolutely same principle. AD optimization used to involve team as
people constantly tweaking campaigns, running ab tests, analyzing results. Now
AI tools can automate that entire feedback loop. They adjust adbids, targeting,
and creative elements in real time based on performance data,
needing far less human oversight.
Speaker 1 (16:51):
And for prime video, it's the recommendation engine.
Speaker 2 (16:54):
Yeah, those what to watch next suggestions, they're constantly being
refined by self learning algorithms. The AI optimizes itself, reducing
the need for large teams of human analysts who used
to manually fine tune those content recommendations.
Speaker 1 (17:09):
Okay, and finally, even Amazon Web Services AWS, the powerhouse,
the profit engine, the foundation for all this AI stuff.
Even they weren't totally immune.
Speaker 2 (17:21):
Even AWS sawcuts specifically in routine cloud support roles. Now,
to be clear, AWS is aggressively hiring top tier AI specialists,
data scientists, engineers building the next generation of services.
Speaker 1 (17:33):
But cutting the basic support right.
Speaker 2 (17:35):
Those frontline roles handling common technical queries from customers. Increasingly
automated advanced chatbots and AI agents can handle a growing
percentage of those routine support tickets, diagnose common problems, and
guide users through fixes.
Speaker 1 (17:48):
So trim the fat on the basic stuff to free
up budget for the high end AI talent.
Speaker 2 (17:52):
That seems to be the strategy across the board. Automate
the routine to invest in the complex.
Speaker 1 (17:57):
So, looking across these areas HR admin, device consideration, demand forecasting,
AD tuning, basic tech support, what's the common thread? What
really defines an AI exposed job? Right now for you listening.
Speaker 2 (18:10):
The common denominator is routine cognitive labor. It's work that
involves processing information, following rules, identifying patterns, and making predictions
based on data, but in a way that's largely repetitive
or standardized.
Speaker 1 (18:22):
And experts agree on this.
Speaker 2 (18:23):
Yes, Analysis from people like Carl Fred Oxford University confirms this.
These are the classic roles vulnerable to AI. Clerical tasks,
customer service interactions, basic data entry and analysis, even entry
level coding that involves generating fairly standard blocks of code.
Speaker 1 (18:39):
You mentioned large language models lms. How are they different
from the algorithms we've had for years? Why is this
wave of AI so potent against these jobs?
Speaker 2 (18:48):
It's the generative aspect. Older algorithms mostly followed pre defined rules.
If this, then do that. Llms can understand natural language,
synthesize information from fast data sets, and generate new con
to emails, reports, codes, summaries, conversational responses that seems human like,
they'll just fall with script exactly. They can handle nuance,
manage multi step tasks, access and summarize information from different sources.
(19:12):
That's why research from places like Stanford shows that hiring
for entry level cognitive roles dropped what was it thirteen percent?
Speaker 1 (19:19):
Yeah, thirteen percent since lms became widespread.
Speaker 2 (19:22):
Because the llms can now do a chunk of that
entry level work.
Speaker 1 (19:25):
So Jasse's push for scrappier teams, it sounds like AI
is the muscle making that happen.
Speaker 2 (19:32):
It's the technological enabler. Absolutely flattening Hierarchyes, becomes much easier
when AI can take over some of the oversight and
reporting functions previously done by junior or mid level manager.
So well, if AI can track per progress, summarize team communications,
flag potential issues, and even delegate routine tasks automatically, then
one human manager can effectively oversee a larger team. The
(19:55):
AI handles the low level administrative oversight, freeing the human
manager for more strategic thinking, or simply allowing the company
to have fewer managers overall.
Speaker 1 (20:04):
Okay, but let's balance this. Amidst these fourteen thousand cuts,
where is Amazon spending money and actively hiring people. They're
not just cutting.
Speaker 2 (20:12):
Yeah, no, they're definitely hiring, but very strategically. They're focusing
on what they call key strategic areas. These are roles
that require deep specialized knowledge, complex problem solving, creativity, things
currently well beyond AI's capability. Advanced AI research and engineering,
sophisticated data science, specialized robotics development, and maintenance. Think about
(20:34):
that big UK investment you hear about forty billion over
three years that promises thousands of jobs. Yes, but they're
largely in building out fulfillment centers and the high tech
AI infrastructure itself.
Speaker 1 (20:45):
So jobs are being created, but they're different kinds of jobs.
Speaker 2 (20:48):
Very different. And the sources emphasize the overall outcome is
still aimed at creating a leaner, meaner Amazon, highly optimized,
highly automated. Even with new hiring in some areas, the
total the whole number of corporate workers relative to revenue
is expected to decrease significantly over time. And it's crucial
to understand this isn't just an Amazon story. What they're
(21:08):
doing is basically writing the playbook for the rest of
the industry, maybe the whole economy. We need to look
at the immediate ripples, the fallout, and the feeling inside
Amazon when this happened.
Speaker 1 (21:18):
Yeah, the sources described a really tense atmosphere, people walking
on pins and needles, waiting for that email to land,
and getting the news via personal email, sometimes at dawn.
That sounds incredibly impersonal and stressful.
Speaker 2 (21:32):
Deeply unsettling, and it creates this profound cognitive dissonance, especially
for the tech workers themselves. There was that quote, I
think from an HR specialist speaking anonymously. Oh yeah, the
paradox of the builder exactly, We're building the tools that
replace us. Imagine spending your career perfecting a technology only
to see it turned around and use to eliminate the
very function you perform. That's a fundamental challenge to the
(21:55):
old social contract between employer and employee.
Speaker 1 (21:58):
And this isn't happening at a vacuum. The tech sector was
already seeing significant layoffs in twenty twenty five even before
this Amazon news hit.
Speaker 2 (22:05):
Oh yeah, the trend was already accelerating layoffs. DOTIFYI, which
tracks this stuff, reported over ninety eight thousand tech job
cuts globally in twenty twenty five prior to Amazon's October announcement.
Amazon is often the bellweather, but they're definitely not alone.
Speaker 1 (22:20):
Who else was making similar moves explicitly citing AI.
Speaker 2 (22:23):
Well. Microsoft shed fifteen thousand rolls in the summer of
twenty twenty five, salesforce cut four thousand in September, and
their executives made that really striking.
Speaker 1 (22:31):
Claim, the one about AI doing half their work.
Speaker 2 (22:34):
Yeah, claiming AI now handles fifty percent of their operational tasks.
Think about what that implies for sales, marketing, customer support roles.
Speaker 1 (22:40):
What does that mean practically? If AI is doing fifty
percent of the work at salesforce, it means.
Speaker 2 (22:45):
The routine stuff drafting initial emails, researching leads, analyzing customal data,
summarizing calls, maybe even generating first draft proposals is increasingly automated.
The human salesperson shifts towards higher level tasks relationship building,
complex negotiation, closing deals, strategic account management.
Speaker 1 (23:04):
Less admin, more strategy, but needing fewer people overall to
hit the same target.
Speaker 2 (23:09):
That's the inevitable consequence. And it's not just software companies.
Look at ups acts fourteen thousand managers partly attributed to
AI optimizing routes and logistics. Luftanza cut four thousand jobs,
citing AI efficiencies and planning and scheduling. It's spreading.
Speaker 1 (23:23):
So there's this interesting economic picture. The sources mentioned that
in September twenty twenty five, the overall US job market
looked okay, adding two hundred and fifty four thousand jobs,
but unemployments still ticked up slightly to four point three percent.
How does that work?
Speaker 2 (23:35):
Yeah, Axios pointed this out. They called it the AI
labor market squeeze. Even if the broader economy is adding
jobs maybe in services, construction, healthcare, the big white collar employers,
the tech giants like Amazon, Microsoft, Google, they're realizing they
don't need to hire as aggressively for those traditional entry
level or mid level corporate roles.
Speaker 1 (23:56):
Anymore because AI is filling the productivity gap or.
Speaker 2 (24:00):
Handling the tasks those roles used to do. So you
get this weird situation. Overall job growth, but a tightening
or even shrinking in the specific sectors that used to
absorb lots of college graduates into cognitive work. The traditional
pathways are narrowing.
Speaker 1 (24:14):
Which brings us squarely to the big debate job destruction
versus job creation. The World Economic Forums twenty twenty five
survey had some pretty dramatic numbers on this.
Speaker 2 (24:23):
The numbers are stark, definitely on the destruction side. The
projection was ninety two million jobs lost globally to AI
by twenty thirty.
Speaker 1 (24:30):
Ninety two million. And these are the kinds of jobs
we've been talking.
Speaker 2 (24:33):
About exactly, clerical work, administrative support, customer service. They projected
chatbots could handle over fifty percent of queries. Basic coding,
routine data analysis, even some layers of middle management focused
on oversight. Anything fundamentally based on repetitive information processing is
at risk.
Speaker 1 (24:51):
But there's always the flip side, right, the argument that
technology creates new jobs.
Speaker 2 (24:55):
That's the counter narrative, and the WEF data supports that too,
at least numerical. They predicted the creation of one hundred
and seventy million new jobs by twenty.
Speaker 1 (25:04):
Thirty, more created than destroyed.
Speaker 2 (25:06):
Then mathematically yes, but these new jobs are very different.
They're in fields like AI ethics, data curation essentially teaching
and refining the AI models, prompt engineering, designing human AI
collaboration workflows, AI system maintenance, highly specialized roles.
Speaker 1 (25:23):
So the net number might be positive, but the skills
don't match up easily.
Speaker 2 (25:26):
That's the massive challenge, the transition risk. It's not a
simple swap. The skills needed for those one hundred and
seventy million new jobs are often vastly different from the
skills possessed by the ninety two million people whose routine
jobs are disappearing, and.
Speaker 1 (25:38):
Who gets hit hardest during this transition. The Burning Glass
Institute had a warning about this being a long process.
Speaker 2 (25:45):
They warn't about a multi decade shakeup, and the data
suggests it's hitting younger workers those just starting out particularly hard.
Right now, Remember that thirteen percent drop and hiring for
entry level coders right that's because the tasks usually given
to junior employees are often the easiest ones to automate
with current AI tools. The challenge is that our current
(26:06):
education systems and corporate training programs are struggling to upskill
people fast enough to meet the demand for these new,
complex AI related roles. There's a growing mismatch.
Speaker 1 (26:16):
This feels like it has really serious social and ethical implications,
especially the difference between this wave of automation and past ones.
Speaker 2 (26:24):
It's a critical distinction. Historically, automation primarily hit blue collar
manual labor in factories, while disruptive society could argue those
jobs were often physically demanding or repetitive. AI is different
because it's targeting cognitive, white collar work, the.
Speaker 1 (26:39):
Jobs that build the modern middle class exactly.
Speaker 2 (26:42):
The analysts, the supervisor, the office worker, the entry level
professional roles. These provided pathways to stable careers, and ironically,
many low skill, physically demanding jobs like warehouse work seem
more resilient. For now, Amazon still needs two hundred and
fifty thousand seasonal hires for packing boxes during the holidays.
Speaker 1 (27:01):
So the corporate ladder is changing fundamentally. You need AI
fluency now to even get on it or stay.
Speaker 2 (27:07):
On it, it seems that way, which risks increasing inequality.
If the pathway to stable middle class corporate work now
requires advanced AI skills that many people lack access to
or can't quickly acquire, you could see a further stratification
of the workforce.
Speaker 1 (27:22):
That warning from five or CEO Mikah Kaufman comes to
mind that AI is coming for everyone. Programmer, lawyer, salesperson.
No white collar role based on routine information processing seems
entirely safe.
Speaker 2 (27:33):
That's the paradigm shift Amazon is helping to accelerate. High
pay or a fancy title doesn't offer protection if the
core function of your job involves tasks that an AI
can perform more efficiently or cheaply. And this forces huge
questions onto policymakers like what like do we need to
seriously consider things like universal basic income? What kind of
massive government supported retraining programs are needed? How do we
(27:56):
ensure the benefits of AI productivity gains are shared broadly
captured by capital owners. These policy debates are lagging way
behind the technological reality Amazon is creating.
Speaker 1 (28:07):
So given all this disruption, what is Amazon actually doing
for the fourteen thousand people they let go? In October
they talked about transition support? What does that practically mean?
Speaker 2 (28:16):
Well, they're offering a package. It includes a period I
think ninety days for affected employees to try and find
another role within Amazon if possible, plus severance pay, access
to outplacement services to help find external jobs, and extended
health benefits.
Speaker 1 (28:30):
Is there any data on how effective that support was
after the earlier twenty twenty two to twenty three cuts.
Speaker 2 (28:36):
The data they released suggests it was reasonably effective relatively speaking.
They reported that around seventy five percent of those laid
off back then either found new roles or left the
company with financial packages they accepted.
Speaker 1 (28:47):
Okay, and what about upskilling for the remaining workforce? Are
they investing there?
Speaker 2 (28:51):
Yes, they're making a significant investment internally. The figure mentioned
is two point five billion dollars specifically for upskilling initiatives.
The stated goal is ambition to help prepare fifty million
people globally, not just employees for working in the AI era,
by twenty thirty.
Speaker 1 (29:07):
Fifty million people. That's huge.
Speaker 2 (29:08):
It's a massive goal. And Jasse's advice to his own
remaining employees was very direct. Educate yourself, experiment with AI.
The message is clear, adapt or risk becoming obsolete. Hashtag
tax tag outro Okay.
Speaker 1 (29:22):
So let's try to summarize the math here. Amazon's fourteen
thousand cuts in October twenty twenty five weren't just random downsizing.
They were a direct, calculated consequence of a strategy to
optimize ruthlessly for AI efficiency and finally deliver on Andy
Jasse's long term goal of slashing bureaucracy. They're actively building
that nimbler.
Speaker 2 (29:41):
Giant exactly, and in doing so, that nimbler giant is
essentially validating the fear that's been lurking in the background
for years. It confirms that yes, AI is a powerful
tool for enabling large scale workforce reductions, all justified in
the name of efficiency and innovation and shareholder value.
Speaker 1 (29:55):
It sets a precedent.
Speaker 2 (29:56):
It absolutely sets a precedent for the entire white collar economy.
When a company like Amazon pouring one hundred billion dollars
into AI infrastructure simultaneously cuts thousands of corporate jobs explicitly
because of AI's potential. That sends a powerful signal. It
forces those crucial policy questions about worker safety nets, retraining mandates,
(30:16):
the future of work itself, questions that frankly, we don't
have good answers for yet.
Speaker 1 (30:20):
So the final takeaway for you listening today seems pretty stark.
This shift isn't coming, it's here. It demands urgent adaptation.
It means actively moving beyond easily automated routine tasks. It
means learning things like trompt engineering, figuring out how to
use AI tools as collaborators, not competitors, and really focusing
on developing those skills that remain uniquely human. Deep critical thinking,
(30:44):
genuine creativity, complex problem solving, and sophisticated emotional intelligence and empathy.
Speaker 2 (30:49):
Couldn't agree more. We know Amazon itself will be reinvented
that internal projection aiming to replace six hundred thousand rolls
by twenty thirty three. That seals Amazon's future trends information.
But the really critical question, the provocative thought may be
for you to take away builds on Jasse's own words.
He said those who embrace the change will be well positioned.
(31:09):
But given the sheer scale of automation planned the systemic
pressure this playbook puts on every company, what happens to
everyone who doesn't get retrained effectively or quickly enough as
this model becomes standard practice, Who inevitably gets left behind
in this AI driven restructuring of the economy. That's the
profound challenge we're all facing now
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com