Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:01):
The AI bubble could potentiallybe worse than the.com bubble, at
least according to one expert openAI turns to Google for its hosting.
Broadcom takes a second shotat making partners hate them.
Delta adopts a newpricing formula using ai,
And guess what?
It's not in your favour.
And Canada's internet registry hassome key insights on everything from
(00:25):
AI adoption to purchasing habits.
Welcome to Hashtag Trending.
I'm your host, Jim Love.
Let's get into it.
A new report from a Top Wall Streeteconomist warns that the current
stock market rally driven by AIoptimism is more extreme than the
peak of the 1990 nine.com bubble.
Torsten Slok a chief economistat Apollo Global Management said
(00:49):
investor expectations aroundartificial intelligence have inflated
the market to dangerous levels.
His data shows that the priceto earnings ratios or PE ratios.
Of the top 10 stocks today areeven higher than they were at the
height of the.com mania in 2000.
Slok writes in 2025, the PE ratiosof the top 10 companies are even
(01:13):
higher than they were at the absolutepeak of the.com bubble in 2000.
And those top 10 firms, mostly largecap tech giants have become the
main engine of the US stock market.
Their gains are masking weakerperformance elsewhere, and their
valuations are being driven by unprovenhopes about AI's economic impact.
(01:35):
At least according to Slok Investorsare treating AI as if it's already
delivered a productivity revolutionand unlocked trillions in savings.
In reality, most of those benefits haven'tmaterialized yet, as Slack puts it.
In 1999, the internet was real.
It did change everything, but thatfact did not stop investors from
(01:56):
wildly overpaying for companiesthat could not deliver on the hype.
He warns that the current marketis fragile, not because AI isn't
real, but because expectationshave outpaced results, and
the rally depends on continuedperfection from a handful of firms.
If history teaches us anything Slackwrites, it's that bubbles do not pop
(02:20):
because their technology is fake.
They pop when investor expectationsdramatically outpace reality.
And with so much riding on a fewovervalued tech giants, a small
stumble could shake the entire market.
if it weren't for the fact thatI'd take a copyright strike on
YouTube I'd be playing BB King's.
(02:41):
The thrill is gone asthe intro to this story.
Open AI has started running coreparts of ChatGPT on Google Cloud.
You heard me right?
Not on Microsoft Azure.
more evidence of how big the Gulf isbecoming between the former partners.
Despite Microsoft's multi-billiondollar investment and deep Azure
(03:03):
integration, OpenAI is now usingGoogle Cloud's custom chips to
power part of chat infrastructure.
Google confirmed the deal on Tuesday
Saying OpenAI is using its TPUV5p chips, which are optimized
for large AI workloads.
This is the first time OpenAI has publiclyacknowledged running major workloads
(03:25):
outside of Microsoft's Azure platform.
The shift comes as OpenAI seeksmore computing power to support
chat GPT and its enterpriseofferings, a demand that may be out
pacing Microsoft's infrastructure, butit also follows recent signs of a strain
between the two firms, including reportsthat Microsoft was caught off guard
(03:46):
by open AI's ChatGPT Search Launch,which competes with its search engine.
Bing.
Google, once seen as a directrival to OpenAI, is now a supplier
suggesting OpenAI is willing topartner wherever it can to get
high performance chips and scale.
Neither company disclosed the sizeof the deal, but Google said open
(04:08):
AI's workloads are already runningin production on its hardware.
This shift sends a clear signal OpenAI is no longer locked into Microsoft's
ecosystem, but it's more than that.
By selecting Google's infrastructureOpenAI is providing an endorsement
of one of Microsoft's biggestcompetitors, which is even more of an
(04:30):
issue because it means that Google hassomething valuable enough that Open
AI would embrace its own competitor.
That's gotta hurt.
But with growth, depending onscarce compute infrastructure,
loyalty may now take a backseatto availability and performance.
Watch this space.
Like any relationship gone sour, there'sa potential for further drama here.
(04:56):
Broadcom is rebooting VMware's partnerprogram for the second time in eight
months, sparking fresh frustrationfrom resellers and integrators
already reeling from previous changes.
The new invitation only Partners programannounced this week limits access to
a smaller group of select resellers.
(05:17):
It replaces the earlier Advantage programintroduced in December, which itself
had caused chaos by abruptly terminatingthousands of partner contracts.
Broadcom said, the new inviteonly model is designed to simplify
operations and drive sales.
Through smaller numbers of strategicpartners, but many in the channel see
(05:39):
it as yet another move to cut them off.
As one consultant told the Register,we were invited into the last new
partner program and now we're out again.
The sudden change left.
Smaller partners uncertain abouttheir status with no clear process
to apply or qualify for the new tier.
Some say that they're now consideringalternative platforms like Nutanix
(06:02):
or Hyper-V to avoid repeateddisruptions to their business.
This comes on top of otherunpopular moves, like the end of
perpetual licenses and VMware'sshift to subscription only pricing.
Broadcom's ongoing reshaping ofVMware looks less like streamlining
and more like scorched earth
With shrinking partner trust and anarrowing field of survivors, with
(06:25):
every reset, more of the channelstarts to look for the exits.
Delta Airlines is reportedly workingon a new AI driven pricing model that
could end fixed fares and replacethem with prices tailored to what
each customer is likely to pay.
According to Fortune, Delta's executivessay they wanna move away from set ticket
(06:48):
prices and towards a system that usesartificial intelligence to determine
how much each individual customer shouldpay based on their browsing, booking
patterns, and willingness to spend.
The goal is to extract more revenueper seat by predicting the maximum
price each flyer will accept a practiceknown as dynamic personalized pricing,
(07:10):
Critics argue this is algorithmicprice discrimination, a model that's
already caused controversy in the rideshare industry where prices can vary
dramatically for the same service.
Delta hasn't confirmed a launch timeline,but its CEO has said the airline is
investing heavily in data science todrive customer centric innovation.
(07:33):
And this could mean two people on thesame flight booking at the same time
might be offered radically differentprices with no clear explanation.
Consumer advocates warn this shift couldmake fair comparisons nearly impossible
and could disproportionately affectlower income or less savvy travelers.
(07:54):
If AI driven pricing becomes thenorm in air travel, the idea of fair
market price could vanish Replaced bya system where what you pay depends
not on cost but on how much thealgorithm thinks it can get from you.
A new report from the Canadian Internet.
Registration Authority shows Canadians areembracing generative ai, But they remain
(08:15):
deeply worried about misinformation,deep fakes, and online safety.
It also reinforces that the Canadianphrase, elbows up, has gone Digital.
Canadians vastly prefer to purchasefrom Canadian online retailers.
The 2025 edition of CIRA's State of theInternet Report reveals that the use of
generative AI has more than doubled inthe past year with one in three Canadians
(08:39):
now saying they've tried AI tools.
Despite that, 74% are still worriedabout AI generated fake content.
Some of the key points, publictrust in X continues to erode.
It now leads all platforms in beingassociated with misinformation
and polarizing content.
(08:59):
Perceived safety on X hasdropped 20 points since 2018.
Even so, only 5% of Canadians haveswitched to alternatives like Blue Sky.
Deep fakes are spreading.
One in three Canadians said they'veencountered a deep fake in the
past year, up from 20% in 2024.
Most Canadians want action.
(09:21):
Nearly 80% of respondents supportbanning deep fakes from social
media and 59% see them as a threatto our democratic elections.
Cyber attacks are personal.
One in five Canadians say they'vebeen the victim of a data breach,
mostly through companies theyuse and elbows are up digitally.
(09:41):
64% prefer to shop from Canadian retailersonline, and over half say the.ca domain
signals a trusted local identity.
Even as Canadians explore newtools like generative AI the
findings show a populationincreasingly wary of digital risk.
(10:01):
The full SIRA report is available@sra.ca.
There's a link in the shownotes@technewsday.ca or.com.
Take your pick.
And that's our show.
Project Synapse is back this weekendwith a lot to talk about in what's been
happening in AI over the past two weeks.
Catch me, Marcel and John onSaturday morning or whenever you
(10:24):
listen to long form podcasts.
I'm your host, Jim Love.
Have a fabulous Friday.