Shownotes
BCG’s Daniel Kuepper and Laura Juliano explain why AI is rewriting the economics of where and how the world manufactures.
Daniel Kuepper, who co-leads BCG's Manufacturing and Physical AI team, and Laura Juliano, BCG's North America Operations Practice Lead, unpack why the factory of the future is no longer a distant vision. Physical AI, agentic systems, and a shared data backbone are converging to expand what's automatable and changing the math on where companies should manufacture. Do leaders still need to separate where to produce from how to produce?
You'll Learn:
What actually separates a genuine “lights-out factory” from a partially automated one.
The biggest mistakes leaders make when starting an AI-led manufacturing transformation.
How new hires are showing up already fluent in AI, changing what factory training even looks like.
Learn More:
BCG’s Latest Thinking on Manufacturing: https://on.bcg.com/46rlzRT
How the Factory of the Future Is Reshaping the Economics of Manufacturing Competitiveness: https://on.bcg.com/4y7h7DW
How Physical AI Is Reshaping Robotics Today—and What Comes Next: https://on.bcg.com/4rgBawW
The CEO’s Guide to Physical AI: https://on.bcg.com/4AoTg4k
Chapters
(00:00) Is Factory of the Future Here?
(01:20) What Is Different This Time?
(04:19) How Significant Are Recent Shifts?
(04:49) Are Lights-Out Factories Realistic?
(07:54) What Industries Are Leading in Robotics?
(09:09) Where Should Factories Double Down?
(11:58) Physical AI's Impact on Companies
(14:55) Leaders’ Costliest Mistakes in Robotics
(18:53) The First-Mover Factory Edge
(21:15) What Should Manufacturing Leaders Do Now?
Meet the Experts
Daniel Kuepper, Managing Director and Senior Partner: https://on.bcg.com/4Am2e27
Laura Juliano, Managing Director and Senior Partner: https://on.bcg.com/4AlXp8Z
Listen to Other Episodes of The So What from BCG podcast
YouTube | https://youtube.com/playlist?list=PLMJgyXjV5gMI9JV-GcF_D1Y6zyf1Eab_0&si=plXqe7-YNzbG56U8
Apple | https://podcasts.apple.com/us/podcast/the-so-what-from-bcg/id1591194141
Spotify | https://open.spotify.com/show/2NSVR7qrAyZ4CaGsnknbBk?si=1d846c2af8784923
Other platforms | https://lnk.to/so-what-general-show12
Follow BCG
LinkedIn | https://www.linkedin.com/company/boston-consulting-group
This podcast uses the following third-party services for analysis:
Podtrac - https://analytics.podtrac.com/privacy-policy-gdrp
Transcripts
- We've been talking
Speaker:about the factory of the
future for a long time.
Speaker:It's been decades down the road,
Speaker:but now it's days down the road.
Speaker:Some industries are actually implementing
Speaker:these technologies
and changes today
Speaker:and seeing real value from them.
Speaker:Others need to start thinking about it.
Speaker:- The real constraint is of
not the technology itself,
Speaker:right, and its maturity.
Speaker:It has been evolving so fast
Speaker:over the last two
to three years.
Speaker:It is rather whether companies are ready
Speaker:to update their ambition
and act on it
Speaker:to ensure future competitiveness.
Speaker:- Welcome to "The So What
from BCG," the podcast
Speaker:that explores the big
ideas shaping business,
Speaker:the economy, and society.
Speaker:I'm Georgie Frost.
Speaker:For decades, companies have
optimized manufacturing
Speaker:around low-cost labor.
Speaker:Now, advances in AI could change all that.
Speaker:The implications go far
beyond the factory floor
Speaker:with the potential to
reshape supply chains,
Speaker:global competitiveness,
and the future of work.
Speaker:Joining me today are
Daniel Kuepper,
Speaker:leading BCG's manufacturing
and physical AI team,
Speaker:and Laura Juliano,
Speaker:BCG's North America
Operations practice lead.
Speaker:Daniel, Laura, welcome to you both.
Speaker:Daniel, we've been talking,
it feels, about the factory
Speaker:of the future and Industry
4.0 for many years now.
Speaker:What is different about this moment?
Speaker:- So for me, and that's
the difference,
Speaker:three key tech evolutions.
Speaker:Number one, physical AI
expands what we can automate.
Speaker:Number two, agentic systems change
Speaker:how the factory is orchestrated,
Speaker:and the scalable technology
backbone changes how quickly
Speaker:successful solutions
can be deployed.
Speaker:- And I'd add one more:
Speaker:talent and the
generational differences
Speaker:of the frontline workers.
Speaker:- We have been talking about
a factory of the future
Speaker:and Industry 4.0 for many years.
Speaker:The ambition was always clear:
Speaker:factories that are more
productive, more flexible,
Speaker:and increasingly self-controlled.
Speaker:What feels different now is
that the technology has
Speaker:finally caught up
with the ambition.
Speaker:Today, I see three things coming together.
Speaker:First, physical AI
can now sense reason
Speaker:and act through robots, machines,
Speaker:and material handling systems.
Speaker:That is a big shift.
Speaker:Robots can adapt and learn in ways
Speaker:that were not possible before.
Speaker:Because many of these systems
can be trained virtually
Speaker:before they are installed
on the shop floor,
Speaker:the economics are improving;
Speaker:the business cases are improving.
Speaker:In our work, we see
Speaker:that roughly 50% more
work can be automated
Speaker:compared with just three years ago.
Speaker:And robot training
and setup effort
Speaker:can be reduced significantly,
Speaker:in some cases by roughly 70%.
Speaker:Second, agentic systems.
Speaker:We are no longer talking only
Speaker:about AI optimizing one
machine or one task.
Speaker:We are talking about AI helping
Speaker:to orchestrate the entire factory.
Speaker:For example, one agent
might continuously optimize
Speaker:the production schedule
Speaker:based on incoming orders and constraints.
Speaker:While another monitors
machine performance,
Speaker:they feed information back to each other,
Speaker:so the factory can
continuously adjust.
Speaker:And then finally, the third topic:
Speaker:We now have the technology
backbone to scale.
Speaker:Technologies like unified
namespace, modern platforms,
Speaker:edge computing, standardized connectivity
Speaker:all play a role in this.
Speaker:In simple terms,
we are finally able
Speaker:to create a common data
layer for the factory.
Speaker:- We're getting to a point now
Speaker:where new graduates and the
entry level of organizations,
Speaker:the machine operators,
Speaker:the folks that actually have
to be driving the back office
Speaker:are coming in with a base
level of understanding
Speaker:of how to interact with
an agent or a chatbot,
Speaker:what it means to have a
machine learning algorithm
Speaker:that underpins the work that is happening,
Speaker:how a robot or basic
physical automation
Speaker:interacts with humans.
Speaker:It is drastically different
to the starting point
Speaker:of where Daniel and I were.
Speaker:I won't say how long ago,
Speaker:but it was a lot longer
than a couple of years.
Speaker:The learning had to happen
immediately on the jobsite
Speaker:versus now that base
is there to build from.
Speaker:- Laura, how significant a shift is that?
Speaker:- Fairly significant.
Speaker:And certainly, the
progress over time has been
Speaker:in fits and starts.
Speaker:Over the last 10 years,
Speaker:I would venture to say 80% of the progress
Speaker:towards physical
automation has been
Speaker:in the last two
or three years.
Speaker:The factors that Daniel just mentioned are
Speaker:really increasing the
slope of the curve
Speaker:at a very rapid pace.
Speaker:- Daniel, Laura there talks
about humans and workers.
Speaker:I'm hearing a lot
Speaker:about so-called
lights-out factories,
Speaker:which doesn't sound as
though it has a lot of room
Speaker:for humans involved unless
they want to work in the dark.
Speaker:What exactly are these things,
Speaker:and why are they getting
so much attention right now?
Speaker:- I would say lights-out
factories are
Speaker:the extreme version of
a factory of the future.
Speaker:It is,
Speaker:it is basically a factory
that tries to operate
Speaker:with significantly
less human operators.
Speaker:Yeah, and that is possible in some places.
Speaker:My personal definition
of a lights-out factory is
Speaker:that we still have indirect
manufacturing labor.
Speaker:We still need maintenance
technicians, yeah, to support
Speaker:whenever there is a breakdown,
Speaker:a physical breakdown
to certain equipment.
Speaker:- Laura, is that vision realistic
Speaker:across manufacturing?
Speaker:- It certainly depends,
Speaker:but I'd say the vision itself
holds for the most part.
Speaker:What it really boils down to is
Speaker:how automated can a process be--
Speaker:whatever the physical process
is that needs to happen
Speaker:to create whatever good is
coming out the back end--
Speaker:and how much do
logistics cost matter
Speaker:in the calculation
of the business case?
Speaker:In some instances,
Speaker:there are some sectors
in industries out there
Speaker:where the, the actual
moves that need to get made
Speaker:or the volume that gets pushed
through a factory is
Speaker:so small or so specialized
Speaker:that the creation
of the technology
Speaker:to have physical AI and
robots conduct it is
Speaker:still a ways away from
a development perspective.
Speaker:In other cases,
it's a financial issue.
Speaker:The business case
just doesn't close.
Speaker:The investment that it would take in order
Speaker:to create a lights-out
factory does not outweigh
Speaker:from a benefit perspective
Speaker:because say there's a
local-for-local setup
Speaker:where everything that
you are making is
Speaker:proximate immediately
to the customer
Speaker:that you're delivering it to.
Speaker:And therefore, in order to
build a lights-out factory,
Speaker:if you have to get farther away,
Speaker:the logistics costs outweigh that benefit.
Speaker:Or perhaps the technology
is just so expensive.
Speaker:Think of high-reliability
industrial environments
Speaker:like satellites
or other technology
Speaker:or medtech where the human
touch is still so important
Speaker:and modeling that in
an automated fashion is
Speaker:not quite there.
Speaker:Now I say "quite" because I
think everything is moving
Speaker:towards that direction.
Speaker:So as we progress over the next few years,
Speaker:there will be aging
assets in manual factories
Speaker:that make a lot less
sense to replace
Speaker:with other manual
machines versus upgrading
Speaker:to more of a lights-out or
physical automation setup.
Speaker:- You mentioned a number
of industries there, Laura.
Speaker:Are any really,
really far ahead?
Speaker:- Certainly, places around
physical technology
Speaker:and those that are heavy in,
Speaker:in regions outside
of the US actually,
Speaker:in Asia, tend to be the ones
Speaker:that have put in the investment,
Speaker:have the high-volume,
low-mix line speed
Speaker:that are so conducive to
this type of investment.
Speaker:And, Daniel, I'm going to
actually pass it to you
Speaker:because I know that's,
that's the space
Speaker:that you work
day in and day out.
Speaker:- Investing in the factory of
the future, first of all,
Speaker:is always in competition
with just relocating
Speaker:to a lower-cost
location, right?
Speaker:And for me, the rule
of thumb is always
Speaker:if the impact that
you can achieve
Speaker:by investing in the factory
of the future is high,
Speaker:and if you operate in a sector
Speaker:that comes along with
significant logistics cost,
Speaker:yeah, so shipping from a distant location
Speaker:to the customers is expensive,
Speaker:if those two things come together,
Speaker:then that is a,
that is a sweet spot
Speaker:for where it makes sense to double down
Speaker:on implementing the factory of the future.
Speaker:- I'm curious to follow this
thread a little bit further
Speaker:and look at the implications
Speaker:of what AI advancements will do
Speaker:for where companies choose to manufacture.
Speaker:You know, if AI
changes that equation,
Speaker:what does it mean
for, for regions?
Speaker:- There have been significant changes
Speaker:over the last 25 years, yeah?
Speaker:Both the US and Western
Europe have each lost
Speaker:11 percentage points
of their share
Speaker:of global manufacturing
value add,
Speaker:shedding over nine million jobs combined,
Speaker:while Asia, and China in
particular, has gained
Speaker:28 percentage points,
Speaker:representing now 55% of global
manufacturing value add.
Speaker:And without a broad
implementation of AI
Speaker:and the factory of the future
that we are discussing here,
Speaker:the current trajectory
will continue.
Speaker:We expect or we see in Europe
alone more than $1 trillion
Speaker:of manufacturing value at
risk of being relocated,
Speaker:and we see more than
$400, $440 billion
Speaker:at risk of being relocated
from the United States.
Speaker:So AI-powered factories can
Speaker:fundamentally alter this equation.
Speaker:By reducing labor intensity in factories,
Speaker:AI reshapes the cost
structure of production.
Speaker:But there is no,
no universal answer
Speaker:whether AI can reverse
the best option
Speaker:where to produce varies
by sector and location
Speaker:as we have said.
Speaker:Western European chemicals,
just to give you two examples,
Speaker:Western European chemicals
or an automotive manufacturer
Speaker:investing in AI-powered factories
Speaker:can be equally competitive
with China when we account
Speaker:for logistics and the
proximity to customers.
Speaker:But in battery cell manufacturing
Speaker:or smartphone assembly,
for example,
Speaker:cost disadvantages are deep enough
Speaker:that even a full implementation
Speaker:of the factory of the future
leaves a meaningful gap.
Speaker:- Daniel mentioned, in China,
Speaker:there is aggressive moves
towards factory of the future
Speaker:to defend its position in automation
Speaker:as this labor cost advantage
equation starts to shift.
Speaker:South Korea has the highest robot density
Speaker:of manufacturing in the
world and continues to pour
Speaker:technology and research
and development
Speaker:as well as on-the-ground
implementation learnings
Speaker:into what they are doing.
Speaker:Even in the US, we're seeing
that tariffs are starting to
Speaker:further motivate the need
to bring more production
Speaker:on shore where
economically possible,
Speaker:and industries like
automotive are able
Speaker:to make decisions about
their entire company future
Speaker:off the back of this capability.
Speaker:- One thing I wanted to,
we've taken a look at sort
Speaker:of the broader picture here,
you know, global economies.
Speaker:Let's bring it back down to companies.
Speaker:I want to explore where they are currently
Speaker:finding value from
AI and automation.
Speaker:- Let me give a few examples
Speaker:because these are ones that are in place
Speaker:and happening every day
Speaker:right around the corner
from each of us.
Speaker:So industrial equipment:
Speaker:Think pumps and valves,
HVAC-type equipment.
Speaker:These are places
where the combination
Speaker:of automation on the line
plus the advanced utilization
Speaker:of data and technology
infrastructure to predict
Speaker:what maintenance will be necessary
Speaker:to increase the runtime of assets
Speaker:that are on the floor
is allowing
Speaker:not only to reduce total labor,
Speaker:but to most importantly,
reduce the churn from labor
Speaker:that comes from the training
Speaker:that is necessary in these roles
Speaker:and the turnover delays
that get driven
Speaker:when attrition happens
in the workforce.
Speaker:It's taking significant
points off of conversion cost,
Speaker:allowing these companies in
Speaker:what in many cases are a hundred years old
Speaker:and have operated the same way
Speaker:and allowing them to
think very differently
Speaker:about their operating costs.
Speaker:Another space, aerospace
and defense, for example,
Speaker:this is a supply chain
Speaker:that often has to be on
the cutting edge of design--
Speaker:incredibly important and
mission-driven sector.
Speaker:There it's less about
taking labor out
Speaker:because the value
of the individual is
Speaker:so important at a
high-value-add job,
Speaker:but it's more about consistency and yield,
Speaker:reducing nonconformance.
Speaker:And so there's
significant value there
Speaker:that's not on the
reduction of cost
Speaker:but rather the increase of what you can do
Speaker:with what you have.
Speaker:- Western economies with high labor costs,
Speaker:they would typically
try to optimize
Speaker:for labor productivity, right?
Speaker:When you have the same discussion,
Speaker:for example, in India,
Speaker:typically no one ever asks
for labor productivity, right?
Speaker:They are all over
asset productivity
Speaker:and a question of how can
we better utilize our assets
Speaker:and also product qualities
and the predominant theme.
Speaker:How can we make sure
that we kind of get
Speaker:to a better quality, a
more consistent quality
Speaker:to be able to serve international markets?
Speaker:So it really depends on
what to optimize for,
Speaker:but the factory of the future
has exactly that in mind
Speaker:and can benefit
Speaker:much more than just
the cost equation.
Speaker:- Laura, you spend a lot of
time helping manufacturers
Speaker:through these kinds
of transformations.
Speaker:What do you see are perhaps
the biggest mistakes
Speaker:that leaders make when
they get started?
Speaker:- One mistake is looking at it
Speaker:as a technology-only solution
Speaker:and, and situation.
Speaker:I know when a client comes to me
Speaker:and asks for what should
our AI transformation be,
Speaker:that I'm in for a long conversation
Speaker:because my question back is,
Speaker:what problems are you trying to solve?
Speaker:And that's the case in
manufacturing as well.
Speaker:What element of your cost
base are you unhappy with
Speaker:or has increased
significantly recently?
Speaker:Where are you seeing
the biggest constraint
Speaker:when it comes to
labor capability?
Speaker:What is it that is
causing excess effort
Speaker:or rework in your
production process?
Speaker:These are the types of questions that need
Speaker:to get asked instead of
where can I use a robot?
Speaker:The second one is
Speaker:underestimating
the people element.
Speaker:I mean, Georgie, you
mentioned this earlier.
Speaker:We talk a lot about
the technology,
Speaker:yet every time that I do,
Speaker:the people become that
much more important.
Speaker:Yes, the profile is
going to be different
Speaker:of what the labor force
looks like at a company
Speaker:that is highly automated
versus not.
Speaker:But as Daniel mentioned,
things like maintenance,
Speaker:prediction, decision-making
power, these are all elements
Speaker:that still have a
human touch to them.
Speaker:And in this interim period
Speaker:before many companies get
to the lights-out factory,
Speaker:there's going to
be a hybrid setup,
Speaker:and you need to focus on
how do we get, first of all,
Speaker:how do we mine people's
knowledge for the areas
Speaker:that are going to
be most effective
Speaker:to utilize this technology?
Speaker:And then how do we
design a solution
Speaker:that augments and
increases the value
Speaker:of what our labor
force is providing
Speaker:and takes away the
low-value-add areas?
Speaker:And then the third one
I'll mention is
Speaker:a mistake I often get:
Speaker:Okay, well, give
me the playbook.
Speaker:What is the playbook?
It's different for everyone.
Speaker:I mean, obviously, Daniel
and I have gigabytes
Speaker:of playbooks that we can provide,
Speaker:but the question is, again,
Speaker:what is the problem
you're trying to solve?
Speaker:What is the unique setup
Speaker:and situation of
your assets, again,
Speaker:many of which have probably
been around for 50 years,
Speaker:your supply chain, the map
of suppliers you're using,
Speaker:your customer base,
what they need
Speaker:and the service levels
that have to be delivered,
Speaker:and how do we apply the
capabilities to those in a way
Speaker:that creates the perfect map for you?
Speaker:What a German auto plant
requires in this way is going
Speaker:to be very different
Speaker:from what a New Jersey
biopharma line looks like
Speaker:at the end of
a transformation.
Speaker:- Daniel, what does it look
like from your perspective?
Speaker:- I would say,
Speaker:what is becoming more
important in these days is
Speaker:to acknowledge that
the two decisions
Speaker:of where to produce
and how to produce--
Speaker:that means with or
without what we discussed
Speaker:as being the factory
of the future--
Speaker:cannot be separated.
Speaker:Those two questions
belong together,
Speaker:and that is different from
what we have seen in the past.
Speaker:Yeah? And that is something
that decision-makers now,
Speaker:when they discuss their
overall footprint,
Speaker:always need to keep,
keep in mind.
Speaker:And then the question is already
Speaker:how much of that reshoring
is already happening?
Speaker:I would say real reshoring
is still rare, right?
Speaker:So it's a decision not to move production
Speaker:to lower-cost locations.
Speaker:It's simply the more
certain response
Speaker:in the eyes of many
decision-makers,
Speaker:given that we are only
at the very beginning
Speaker:of using the latest
AI advancements
Speaker:and change manufacturing
economics at scale.
Speaker:- Daniel, you say that
we're at the very start.
Speaker:Is there a benefit
to, you know,
Speaker:the first movers in this
space or the watch-and-sees?
Speaker:- There are, I mean, there
are definitely benefits.
Speaker:So I would argue we
are ready to deploy
Speaker:and we are deploying it
for many of our clients.
Speaker:There's still, from my
perspective, though,
Speaker:a gap in the awareness
Speaker:at many manufacturing leaders.
Speaker:So many of them still look at
automation through the lens
Speaker:of what was possible
a few years ago--
Speaker:expensive, rigid, hard to
program, difficult to scale--
Speaker:but some technology
has moved very fast.
Speaker:Physical AI, agentic systems,
Speaker:and scalable data
infrastructure have changed
Speaker:both what is possible and what
is economically attractive.
Speaker:And then sometimes
I also believe
Speaker:there's a lack of courage.
Speaker:Companies need to move
beyond isolated pilots
Speaker:and define a more
ambitious target picture
Speaker:for how the factory
Speaker:and ultimately the factory
network should operate.
Speaker:And certainly I would say,
Speaker:companies need to invest
behind it, not blindly
Speaker:but with conviction
in automation,
Speaker:in the data backbone
and capabilities,
Speaker:and in change on
the shop floor.
Speaker:And yes, this investment
will not always pay off
Speaker:in one or two years.
Speaker:That is clear.
Speaker:So what I would say is
Speaker:the real constraint is of
not the technology itself,
Speaker:right, and its maturity.
Speaker:It has been evolving so fast
Speaker:over the last two
to three years.
Speaker:We discussed that.
Speaker:It is rather whether
companies are ready
Speaker:to update their ambition
and act on it
Speaker:to ensure future
competitiveness.
Speaker:- Laura, courage and conviction.
Speaker:- Absolutely. I mean,
I would say,
Speaker:just to summarize
what Daniel said,
Speaker:moving forward,
leaders need to define
Speaker:what problem they're trying
to solve, set a vision,
Speaker:and then what the first
step is going to be
Speaker:because a vision without
that first actionable step
Speaker:with the investment is never
going to get off the ground.
Speaker:And a first step without
any vision associated
Speaker:is going to stall.
Speaker:- Well, you've led
me beautifully
Speaker:to the next and
final question
Speaker:which is the now what.
Speaker:What is the first step, if
I'm a manufacturing leader,
Speaker:that I should be taking tomorrow?
Speaker:- Well, for me, the first step
tomorrow morning is simple.
Speaker:Update your view of what is now possible,
Speaker:define a bold but
practical factory ambition
Speaker:anchored in your highest
value opportunities,
Speaker:and invest behind it.
Speaker:The factory of the future
is no longer, for me,
Speaker:just a vision or a
collection of pilots.
Speaker:It is becoming an executable
transformation agenda
Speaker:with significant implications
Speaker:for productivity
and competitiveness.
Speaker:- Laura?
Speaker:- Said beautifully, Daniel.
Speaker:I would just add to it,
evaluate your people,
Speaker:listen to your people,
and understand together
Speaker:how you're going to
go on this journey.
Speaker:- Laura, Daniel,
thank you so much.
Speaker:And thank you for listening.
Speaker:If you'd like to find out
more about this subject,
Speaker:you can read the article
Speaker:"How the Factory of
the Future Is
Speaker:Reshaping the Economics
Speaker:of Manufacturing
Competitiveness."
Speaker:You just find the link in the show notes.
