/Tech8h ago

Gary Marcus Agrees AI Boom Is Huge but Calls SoftBank Bet Too Early

1899161411.1K
Gary Marcus#156
Original post
Gary Marcus@GaryMarcus#156inTech

Dude can’t even get a loan on his OpenAI shares …

but I actually agree … AI will be huge …

That said, I think Masa bet too early, and on the wrong horse.

Too much on LLMs; too much on Sam; too early.

Better AI will definitely come.

This generation of AI, however, won’t live up to expectations. The fundamentals aren’t sound.

CNBC@CNBC

BREAKING: The AI revolution will be 50 times bigger than the Dotcom revolution in the 2000s, SoftBank CEO Masayoshi Son told CNBC Monday.

“I think this is like more than 10x, probably 50x bigger than dotcom,” Son told CNBC’s Arjun Kharpal in Paris a day after the company announced that it’s investing 75 billion euros ($87 billion) to build AI infrastructure in France, including 5 GW of AI data center capacity.

More details: http://cnb.cx/4dD9DAQ

9:45 AM · Jun 10, 2026 · 1.8K Views
Sentiment

Many users agree with Gary Marcus that SoftBank's AI bet is too early, while others argue recent inference cost drops make 50x ambitions conservative or dismiss current LLMs as overhyped.

Pos
62.5%
Neg
37.5%
9 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS9.4KBOOKMARKS13LIKES90RETWEETS16REPLIES12
Gary Marcus@GaryMarcus

Masa can hype AI all he likes, but apparently he can’t even get a loan on his OpenAI shares. 🤔

But I actually half agree—and it might surprise some of my critics to hear this, especially those who haven’t listened carefully—with what he said below: AI *will* be huge. Maybe 50x Dotcom. like he says.

But I think Masa bet too early, and on the wrong horse.

Too much on LLMs; too much on Sam; too early.

Better AI will definitely come.

This generation of AI, however, won’t live up to expectations. The fundamentals aren’t sound.

CNBC@CNBC

BREAKING: The AI revolution will be 50 times bigger than the Dotcom revolution in the 2000s, SoftBank CEO Masayoshi Son told CNBC Monday.

“I think this is like more than 10x, probably 50x bigger than dotcom,” Son told CNBC’s Arjun Kharpal in Paris a day after the company announced that it’s investing 75 billion euros ($87 billion) to build AI infrastructure in France, including 5 GW of AI data center capacity.

More details: http://cnb.cx/4dD9DAQ

8hViews 9.4KLikes 90Bookmarks 13
A Eye Bubble@A_Eye_Bubble

@GaryMarcus Yes

8hViews 107Likes 1Bookmarks 1
andreas petz@petz_a

@GaryMarcus Who do they think they can impress with 50x? Jensen is at billion x (inference)

8hViews 174
Ferbin@Ferbin08

@GaryMarcus Disagree that Masa's finances matter here. Also disagree that 50x is ambitious. Inference costs dropped 10x in three years. You can build robotics, voice assistants, smart devices now that you couldn't before. 50x might be conservative.

8hViews 148
keyser noway 🪼@KNohue

@GaryMarcus Technology and tech investing never about competition. It was about monopolies. You get there first and you generally win the monopoly: they call it leveling up.

Zuck did that with the meatheadverse. He got there first, but no one followed. lol

LLMs are not new tech 1/2

6hViews 102
JohnSalmondOz@OzSalmond

@GaryMarcus Reminder that, for all his smart commentary on the insane reality of American billionaire AI, Marcus is still an AI dreamer

3hViews 48
JohnSalmondOz@OzSalmond

@petz_a @GaryMarcus Surely gods are entirely unlimited, and a god is what AI is now depicted as

3hViews 4
hidflect@hidflect

@GaryMarcus Yes. You nailed it.

4hViews 44Likes 1
steven Y@y_steven35383

@GaryMarcus I totally agree with you. Too early and on the wrong horse for Masa!

4hViews 38Likes 1
DesignCntrl Inc. / Destrozado@DesignCntrl

@GaryMarcus The dotcom boom took 10 years to become useful which lead to years of gatekeeping by Google.

5hViews 50
Saquib Mehmood@SaquibOptimusAI

@GaryMarcus You are on a roll bro, for the last couple of days. Hope the sun keeps shining on your world.

4hViews 34
Hunter Gon@gonlenidefi

@GaryMarcus wait so he can borrow against $OPEN but SoftBank stock is fine?

you see the hierarchy right there

8hViews 21
Simon Kim@SimonKi04473385

@Ferbin08 @GaryMarcus Can you point me toward the source about inference costs dropping 10x in the last 3 years? If true, AI is definitely economically viable.

8hViews 15
keyser noway 🪼@KNohue

@GaryMarcus have been hallucinating for 20 years before Altman released ChatGPT.

So glad I listened to a podcast with you four years ago and immediately knew it was all hype.

i’m just a rando no background in tech and i could see through it lol

6hViews 14
chemoshi@chemosh933

@GaryMarcus Yes, real AI, not LLM or the anthro hybrid. These are overglorified search engines..

2hViews 12
fitnfinance@fit_and_finance

@GaryMarcus 100% agree. The infrastructure built now will be the groundwork for the future high-value companies. Short / medium term though the valuations are too high to be achievable.

1hViews 10
Blissy@BlissyOnX

@GaryMarcus 50x bigger but same liquidity issues as private shares

guess the revolution needs a check first

8h
Invincible@InvincibleEdge

@GaryMarcus the margin call on hype timelines is wild

wait so LLMs are the dot com bubble but without the buildings

8h