Hey👋,
I'm Giacomo

Elon never wanted to buy OpenAI.

‍

Instead, this is what his $97.4 billion offer really meant:

‍

🔹 OpenAI is transitioning from non-profit to for-profit.

To do so, a new for-profit entity needs to acquire OpenAI’s assets from the non-profit, ideally at the lowest possible price.

Since OpenAI is still burning billions, cash is a concern. And when you're a private company, you can value your assets however you like (kind of).

$1 inter-company transactions are not unheard of.

‍

🔹 But Elon is messing with Sam's plans.

His formal $97.4B offer put a price tag on OpenAI’s assets.

Now, it’s much harder for Sam Altman to justify a low valuation for the transition.

Higher price means more cash to raise and slower progress.

‍

🔹 Elon's team also committed to keeping OpenAI non-profit,

right when Sam is pivoting it to a for-profit entity.

Not only that, they will even withdraw the offer should OpenAI decided to stop the transition.

The message is clear: Elon is the "good guy", prioritising humanity over profits, unlike Altman.

‍

🔹 Then, the killer move: due diligence.

The offer letter included a May 10th deadline and demanded access to OpenAI’s "assets, facilities, equipment, books, and records" as well as "particular personnel involved in the Business" for interviews.

Even if the deal didn't go through, meanwhile xAI would get invaluable insights into its competitor’s business and technology.

‍

🔹Elon got exactly what he wanted.

Few days ago, OpenAI’s board formally rejected the offer.

But Elon still got what he wanted:

‍

- Undermine Sam Altman's reputation.

- Regulatory scrutiny over the for-profit transition.

- Slowdown in OpenAI operations and plans.

‍

Often dismissed as just erratic and crazy, Elon is actually playing a long and calculated game. Later today he will release the latest xAI models, supposed to be the best of all.

‍

I wouldn't be surprised if Elon eventually won the AI race, same as he did with EVs and space.

‍

Time will tell!

The US is not at the forefront of EVs.

‍

Will this give an edge to Europe at some point?

Forget SEO.

‍

Welcome to AEO (Agent Engine Optimisation).

‍

I tested OpenAI Operator by asking to find a suitable apartment for rent in Zurich within a given budget.

‍

Operator opened its browser.

‍

It used Google Chrome, but it didn’t search on Google.

‍

It searched on Bing (no surprise here, given Microsoft’s massive investment).

‍

Then things got interesting.Instead of heading to Homegate or ImmoScout24 (specialised real estate platforms in Switzerland, owned by my employer), it went straight to comparis.ch AG, a general comparison site and my employer's competitor.

‍

I noticed a similar pattern when searching for flights.

‍

Rather than Skyscanner, it often defaults to Google Flights.

‍

How can you make sure agents choose your website over competitors'?

‍

Does it even matter?

‍

Well, it does.

‍

At the end of the process, which lasted about 10 minutes, Operator returned two links of two suitable apartments. I clicked on them and eventually interacted with the website myself.

‍

💡You want to make sure those links bring to your website!

‍

In a world where agents browse on behalf of humans, SEO and marketing strategies will need to evolve.

‍

UI, UX, copywriting and SEO will have to shift from human-centric to agent-centric design. Remember, agents don't feel emotions! This changes everything.

‍

Are marketers ready for this?

I went all in! 🔥

‍

I subscribed to ChatGPT-Pro for $200/month.

‍

Like someone way smarter than me once said:

"The risk of under-investing is dramatically greater than the risk of over-investing."

— Sundar Pichai, CEO of Alphabet Inc.

‍

And I believe this is exactly the case with #AI.

‍

I've been experimenting with AI tools for over two years, but only now do I feel something fundamentally shifting.

‍

The o1 model, released last autumn, is years ahead of GPT-4o.

It's so good that I kept hitting the usage limit with my Plus subscription.

‍

So, I subscribed to Claude to try their latest 3.5 Sonnet model.

Same issue, kept hitting the limit.

‍

Tried DeepSeek... good quality, when it works.

It’s often unavailable and lacks the polished UX of competitors. No custom GPTs, no Claude Projects etc.

‍

Then, I discovered that GPT-Pro unlocks an even more advanced model: o1-pro, with no usage limits. 🤯

As if that wasn’t enough, OpenAI dropped Deep Search last week, meant to be a complete game changer.

Add Sora and Operator (via VPN), and I was sold.

‍

First impressions: blown away.

‍

Deep Search + o1-pro is the real deal.

In minutes, it completed research tasks I had worked on for months.

‍

Sora seems far from the mind-blowing demos we saw months ago.

I’d say it’s comparable with Runway. But, for $200/month it's virtually unlimited.

Try replicating the same projects with Runway, and you'll easily spend the same amount or more. A single video project I worked on in September costed me around $70. So, all in all Sora is a good deal.

‍

Operator is also a bit underwhelming, but gives a clear idea of where Agents are heading.

‍

Not sure how long I’ll keep the subscription, but for now I’m all in.

‍

Over the next few weeks, I’ll be posting some real-life use cases of GPT-Pro in the context of content and marketing.

‍

Stay tuned!

You make a lot of money, do you?

Sam Altman: "No"

"I'm paid enough for health insurance, I'm doing this because I love it."

‍

Scam Altman 😂

‍

Can't figure if this is all a joke or a computer simulation. Either way, it cannot be real.

#AI use case of the day: make charts out of images 📊

‍

Yesterday I posted some data about Spotify. All charts were made with ChatGPT.

‍

No Excel, no PowerPoint, no copy-pasting numbers.

‍

1. Asked GPT to find Spotify's latest financial report.

2. Took a screenshot of the Annual Financial Summary table

3. Uploaded screenshot in ChatGPT and asked to make a chart out of it.

‍

Sure, it took me a while to chat my way through the final result, it was worth it!

‍

Would have I been faster by just copy-pasting numbers into Excel?

‍

maybe... but what if I was dealing with a more complex table? I guess I would have given up.

‍

AI helps:

‍

-> Make existing tasks faster.

-> Unlock tasks you would have never done otherwise.

If you’re not planning to run audio ads in 2025, you should start now 👇

‍

Young Americans spend over half of their daily listening time with ad-supported digital audio, whether it's podcasts or music streaming. I guess similar picture in Europe.

‍

Spotify reigns supreme.

‍

With 32% of the global streaming market and rapidly growing monthly active users, it’s the king of audio.

‍

In 2024, ad-supported users grew faster than premium ones, now making up 62% of its total user base (vs 56% in 2020).

Advertising revenue are up 7% year-on-year.

‍

Until recently, Spotify relied heavily on direct ad sales. But now, they’re expanding their programmatic solutions in partnership with The Trade Desk.

‍

And this is where the opportunity for brands arises!

‍

Advertisers can now reach 425 million ad-supported users in a programmatic and efficient way.

‍

And here’s the best part:

recording audio ads is cheaper than ever with tools like ElevenLabs and its AI-generated voices.

‍

And if this wasn't enough,

there's a massive gap in the market 💡.

‍

Audio consumption is growing, but it’s still under-monetised.

‍WARC estimates that while audio accounts for 24.5% of all ad-supported media consumption in the US, it represents only 8.4% of ad spend. 🤯

‍

This means less competition, lower costs and more opportunities "to be heard".

‍

Audio-only social media (like Clubhouse) tried and failed.

Spotify took it all. It now owns everything audio, user-generated music, podcasts, even audiobooks.

‍

In this landscape, music becomes "content" and musicians become "creators", just like YouTubers and podcasters.

‍

While this might not be good news for musicians, it is great news for advertisers!

‍

Do you have experience with audio ads?

🚨 Elon Musk offers ~$100 billion to buy OpenAI!

‍

Sam Altman politely declines: "no, thank you"

"but we’ll buy twitter for $9.74 billion if you want."

‍

Two kids taunting each other on a playground. 😂

‍

Elon lowballs OpenAI at less than half its estimated $260B valuation.

‍

Sam reminds him of his $44 billion overpriced Twitter acquisition, now worth less than 1/10.

‍

Elon was right all along…

We do live in a simulation!

NVIDIA’s stock crash last week was a glimpse of the next Black Swan.

‍

But what is a Black Swan? 👇

‍

In business, a Black Swan is a sudden unpredictable event with the power to disrupt an entire market, or even the global economy.

Think of the 2008 financial crisis, COVID or 9/11.

‍

People and markets try to predict and rationalise these events, but most Black Swans are just impossible to foresee. At least in the way they unfold.

‍

Only time will tell if last week’s drop was just a momentary panic or the first sign of something much bigger.

‍

In his later book, Nassim Taleb argues that people and companies can prepare for Black Swan events by developing Antifragility.

‍

Fragile things break under stress (like glass).

Resilient things survive stress but don’t improve (like steel).

‍

Antifragile things thrive under stress (like muscles that grow through exercise).

‍

The ability to iterate fast, change course and prepare plans B, C, D characterise Antifragile people and companies.

‍

💡Antifragility is the #1 skill for the next decade!

‍

Agreed?

Just by looking at demos of OpenAI Deep Search and Operator, I can totally see how AI Agents could kill Big 4 and MBBs.

‍

Unless...

‍

Big 4 and MBB will embrace AI agents to their advantage. Which they'll definitely do.

‍

But, they'll still get loads of competition from individual AI experts that will deliver as good jobs for 1/10 of the fees.

‍

For the first time in history, single-person companies might represent a serious threat to large businesses.

‍

Future is bright for individual creative entrepreneurs!