00:00:00
In the last month or so,
companies involved in the
00:00:04
development of artificial
intelligence have taken a bit of
00:00:06
a hit on Wall Street. There
seems to be some concern amongst
00:00:10
investors about the future and
the impact of AI as a main
00:00:14
component of our culture. It
hasn't stopped the development
00:00:17
of the technology, but it is
kind of an interesting
00:00:20
inflection point as we move
forward. To explain what's going
00:00:24
on and how it may impact in the
future, pleasure to be joined
00:00:27
once again by Wharton Marketing
Professor Stefano Puntoni, who's
00:00:30
also co-director of the Wharton
Human AI Research Program.
00:00:34
Stefano, great to catch up with
you again. How are you, sir?
00:00:37
Hi. Dan. Great to be here. Thank
you.
00:00:39
Thank you. You know,
00:00:41
I'm obviously somebody who
follows the business side of a
00:00:44
lot of these technologies fairly
closely. It has been very
00:00:46
interesting to me to watch from
a shareholder perspective and
00:00:51
see how there has been this
impact on stocks in the last
00:00:55
couple of months because of
these concerns.
00:00:57
Yeah. So in the month of February,
we've seen especially software
00:01:02
stocks being hit quite hard. The
Standard & Poor's software index is
00:01:10
down over 20% since the
beginning of February, and
00:01:14
people have been talking about
the SaaS-pocalypse. So this idea
00:01:17
that many software companies are
going to find it hard to sustain
00:01:23
the margins and businesses they've have
had over the last few decades
00:01:26
moving forward because of AI.
00:01:29
So this is being referred to as a
SaaS-pocalypse, meaning, what, exactly?
00:01:35
Yeah, this idea that
software as a service, or SaaS,
00:01:39
is a kind of business that has
been very attractive, very high
00:01:43
margin, recurring revenues, and
many companies have grown very
00:01:48
large off the back of software
delivered that way. And there
00:01:52
are questions now about the
sustainability of that industry
00:01:57
in the face of increasingly
capable AI agents
00:02:01
and AI systems.
00:02:03
So when you dig deeper, what are
some of the actual concerns that—
00:02:08
that people have right now?
00:02:10
So last week, we were in San
Francisco, and we did a
00:02:14
roundtable with an amazing group
of senior business leaders to
00:02:19
discuss the impact of AI on
value creation. And this is one
00:02:24
of the topics that came up. So I
can share, maybe, a few of the
00:02:29
observations then that our
business leaders made about the
00:02:34
impact of AI on software
companies. There were at least
00:02:39
three strands of that
conversation as to where is the
00:02:45
threat coming from. And one
threat comes from the fact that
00:02:50
this industry presents market
opportunities. Like I said
00:02:54
before, margins have been
historically very high for
00:02:57
decades. Many customers are kind
of locked into the services
00:03:03
provided by the SaaS platforms,
whether those are ERPs or CRM or
00:03:08
lots of different types of
business solutions that
00:03:11
enterprises need to run their
operations. And they're quite
00:03:16
costly. And many customers might
not be delighted. So there's
00:03:19
basically an opportunity there
to disrupt that market. That's
00:03:23
one. The second one is the fact
that building an enterprise
00:03:28
platform is much easier today. I
mean, to create those enterprise-
00:03:33
grade software solutions used to
be extremely difficult and
00:03:38
expensive. It is much easier
today with coding agents, which
00:03:42
basically is lowering barriers
to entry, and so competition
00:03:46
should be expected to be more
aggressive. And then third,
00:03:51
companies everywhere are
rethinking their workflows,
00:03:55
rethinking the way that they get
stuff done across functions and
00:03:59
across verticals. So the idea of a
standardized workflow, which is
00:04:06
really central to this notion of
the SaaS companies, because
00:04:09
basically they sell the same
product to lots of different
00:04:12
companies, might be under
pressure. In particular, what
00:04:16
our roundtable participants were
highlighting was that moving
00:04:20
forward, it may be more valuable
to have very deep expertise in a
00:04:25
vertical rather than having very
deep expertise in how to deploy
00:04:31
a standardized workflow across
verticals. Just to explain a
00:04:34
little better, if I am providing
an enterprise customer with a
00:04:38
software solution specialized on
healthcare, and now healthcare
00:04:43
companies are rethinking lots of
workflows, then maybe that
00:04:46
expertise is more valuable than
having a software that does one
00:04:49
specific workflow in a very
efficient or standardized or
00:04:53
reliable way. And so there might
be opportunities there, almost
00:04:56
for flipping the matrix and
having competitors for enterprise
00:05:00
software products or solutions
being competing more around
00:05:04
expertise revolving particular
issues in an industry, rather
00:05:07
than a specific workflow like a
CRM or whatever. And so that was
00:05:12
what our roundtable participants
were saying. And when you think
00:05:16
about it now, we have the three
forces. You have maybe not
00:05:21
totally satisfied customers, so
opportunities to switch. You've
00:05:25
got a lower barrier to entry.
And then you've got a shifting
00:05:30
in the value propositions sought
after by— in this industry. And
00:05:35
all that together could lead to
undermining the position of many
00:05:40
of these SaaS companies. Now,
our round table participants did
00:05:44
not think that this company
would cease to exist. They
00:05:47
thought that you would still
need the foundation layer, so to
00:05:50
speak, and that they would not
want to have all the data
00:05:54
dependent on one particular AI
solution vendor, whether that is
00:05:57
Anthropic, Open AI, or Google, or
whatever. And so maybe they
00:06:01
would— they would want to
retain, but the margins could be
00:06:04
expected to decrease quite
considerably.
00:06:07
Yeah. And part of this, from what I
understand, is also a discussion
00:06:13
about pricing as well, correct?
00:06:15
Yeah. So there was a
discussion about that, too. At
00:06:18
the center of that has been— in
the media and across companies—
00:06:24
this idea that maybe there's
pressure to shift away from an
00:06:28
input-based pricing model— for
example, the number of users— to
00:06:32
a more output-based pricing
model, which is more like, you
00:06:35
know, the benefit that the
software is supposed to provide
00:06:38
to the company. And some AI
companies are trying to push
00:06:41
this output-based pricing, which
might undermine the
00:06:45
profitability of some of the
SaaS companies, because that
00:06:47
input-based pricing is quite
profitable. And our participants
00:06:52
at the roundtable were a little
skeptical that this would be
00:06:55
happening very fast and across
lots of different software
00:07:00
solutions. The reason being that
to be able to price on outputs,
00:07:05
you need to be able to measure
outputs reliably and in detail.
00:07:10
And for many of those complex
organizational workflows and
00:07:14
organizational outcomes, it's
quite difficult to do. And so in
00:07:17
the case of some of them, like—
I don't know, maybe call centers
00:07:22
or whatever, where you have
clear output metrics,
00:07:24
objectively measured, reliably
measured, trackable over time—
00:07:28
then maybe that is going to
happen, or there'll be pressure
00:07:31
towards that. Maybe it won't
happen, but it will still
00:07:34
decrease the margins, even if
you stick to an input-based
00:07:37
metric. But for some other—
imagine some legal services or
00:07:42
whatever, it might be difficult
to actually move there, because
00:07:44
it might be difficult, for example,
to track the quality of the
00:07:47
output. So we'll have to see.
00:07:49
I think that's a little
bit of an open question, I'd say.
00:07:51
I would think that
presents an interesting, you
00:07:54
know, scenario, when you're
talking about the software
00:07:57
developers versus the companies
that want to use the software,
00:08:01
because the companies that use
it, they obviously want a
00:08:04
dynamic of the number of users.
That's kind of what we've seen
00:08:08
built around technology on the
internet anyway. So does this
00:08:12
present, you know, kind of this
unique dynamic as we look at the
00:08:16
development of this technology
moving forward, as to how both
00:08:21
sides of the technology are
really looking to to move this
00:08:26
forward?
00:08:27
Yeah, there's going to be very
interesting competitive
00:08:29
dynamics. And I'm not a
professor of strategy, I'm a
00:08:32
professor of marketing, so it's
a little bit beyond my expertise.
00:08:34
But you can imagine, you know,
if you go back to your standard
00:08:39
frameworks and competitive
analysis, like Michael Porter's
00:08:42
Five Forces, or something like
that, the arrival of these very
00:08:45
powerful AI agents and the
competition, the convergence of
00:08:49
competition from AI and other
tech sectors into software,
00:08:55
you're going to basically see
changes in dynamic competition.
00:08:59
It's not only the SaaS companies
are feeling the pressure. In many
00:09:02
situations, also, their clients
are feeling the pressure. So
00:09:05
you're going to have competitive
pressures across the value
00:09:08
chain, not just in one
particular area.
00:09:12
So then, with this discussion
going on, is this kind of an
00:09:15
important shift moment when we
think about the importance of
00:09:18
artificial intelligence moving
forward? Or is this, I guess,
00:09:22
just a little bit of a bump in
the road?
00:09:24
It is— I mean, it's impossible
for me to say whether the
00:09:29
depressed value of software
stocks is going to be more of a
00:09:32
permanent feature or something
that will change, but I have to
00:09:35
say that I think that the month
of February was— might, in the
00:09:40
future, might be looked at as a
bit of an inflection point. Up
00:09:44
to few weeks ago, I would say
that AI was a very important
00:09:48
topic, mostly sold on its
potential. What Nate Silver
00:09:52
calls "the white paper phase."
It's like, you know, this is
00:09:55
going to happen, and you ought
to pay attention. And I think
00:09:57
what we've seen over the last
few weeks is going to
00:10:01
be maybe a change where now
we're talking about impact and
00:10:04
not potential. So there have
been some discussions about, you
00:10:07
know, Eric Brynjolfsson from Stanford
arguing that we're starting to
00:10:10
see the impact on productivity
statistics. Controversial, but
00:10:13
maybe we are. We've seen coding
agents and other types of
00:10:16
agentic workflows becoming much
more powerful and more widely
00:10:21
adopted. We've even seen the
massive role that AI is playing
00:10:26
on geopolitics. Just look at the
latest drama around Anthropic
00:10:31
and the Department of War. And
so there's been few signs that
00:10:35
we might be moving in from the
stage where, oh, this is a
00:10:38
powerful technology people ought to
pay attention to, because in the
00:10:42
future it will have an impact to
use the power of technology, and it is
00:10:45
having an impact now.
00:10:47
Stefano, great to have you back
with us again. Thanks very much
00:10:49
for your time today.
00:10:50
Thanks, Dan.
Great to be here.
00:10:52
Thank you. Stefano Puntoni,
00:10:54
marketing professor
here at the Wharton school.
