Boom Or Bust? The AI Bubble Examined

Welcome to another edition of the Neural Net…

It’s Friday. The weekend is practically knocking and even the robots are ready to wrap things up.

In this issue:

• The AI “Bubble” Examined

• Meta Designs In-House AI Chip

• Waymo Gives Google Employees More Time To Work

• Anthropic Not As Independent As Previously Thought

• How France Measures Up To Global AI Funding

• Dream AI Application? A Super Assistant!

Are We In An AI Bubble?

Is AI Riding a Bubble, and Will It Burst?
Some folks think that AI’s current hype is outpacing reality, sparking concerns about a looming “bubble.” On one hand, breakthrough tools like ChatGPT are undeniably game-changing. On the other hand, eye-popping valuations have every startup slapping “AI” on their pitch decks—and it’s actually working. Regardless, its important to keep perspective—if a bust happens, it likely just means weaker players exit while genuinely valuable AI solutions continue to evolve.

So how would we go from sky-high excitement to a possible crash? Let’s take a look at the typical boom-to-bust journey:

1. Spark of Innovation: A major breakthrough (like ChatGPT) ignites everyone’s imagination.

2. Rising Tide: Hype escalates, and investment floods in.

3. Euphoria: Valuations surge to levels that may outpace real earnings.

4. Smart Exits: More cautious investors sell high, predicting a correction.

5. Rapid Downturn: A sell off causes a sudden drop in valuations leaving many with losses.

Hype vs. Reality—How to Spot the Bubble
Historically, bubbles form when enthusiasm outpaces substance, much like the dotcom era burst. It wasn’t that the internet was worthless—it was just overvalued. Massive funding rounds that might not match actual product viability, and price-to-earnings ratios that look unhinged, are all potential bubble indicators. Here are a few more:

• Missed earnings: Big-name AI companies failing to deliver on lofty profit expectations.

• Fading investor and consumer enthusiasm: A new “next best thing” emerges, drawing interest and funds away from AI.

• Technology plateau: Reaching the limits of GPUs, compute, or training data—where tools like ChatGPT stop making noticeable leaps.

• Increased government regulations: New legal constraints that stifle rapid growth or raise compliance costs.

The Key Question
If AI is a bubble, what stage are we in? And should we be concerned? The chart below shows a clear uptake in AI references on corporate earnings calls around the time ChatGPT launched in 2022.

AI has undeniably drawn the attention of big tech and the broader market for the past few years—except for Warren Buffett’s Berkshire Hathaway—due to its transformational potential. For some companies, the mention of AI in earnings calls peaked around 2023, declined, and is now on the rise again. This could suggest we are still in the Rising Tide stage.

Regardless, the sheer level of investment in AI-related ventures is concerning, particularly amid broader economic uncertainty. While we believe AI will change the world, a market correction may be necessary to eliminate excess.

Heard in the Server Room
Quick hits roundup of AI news

• Meta Completes Design Of Secretive In-House AI Chip

  Meta has passed a key design milestone for its in-house AI training chip, aiming to reduce reliance on external suppliers like Nvidia and lower infrastructure costs associated with its AI investments. Partnering with Taiwan's TSMC (Taiwan Saves Mark Cash?), Meta aims to scale up production if tests are successful and expects to integrate the chip into its training platforms by 2026. This would allow Meta to go in-house and not rely on Nvidia’s expensive chips.

• Waymo Hits The Road In Silicon Valley

  Waymo, Google’s self-driving car, is expanding its robotaxi service in the San Francisco Bay Area, adding a 27-square-mile area to its operations. Initially, rides will be available to select Waymo One users, with plans to gradually extend access to more riders. The expansion connects key locations such as multiple Caltrain stations, Google's headquarters, and the Shoreline Amphitheater. Standby for the announcement from Sergey Brin that employees should now work on their commute.

• Google Unmasked As Major Anthropic Stakeholder

  Google's relationship with AI startup Anthropic is much cozier than previously thought. Court documents reveal the search giant owns a 14% stake and has committed over $3 billion total, with another $750 million coming this year. While Google lacks formal control through voting rights or board seats, the substantial financial backing raises eyebrows about Anthropic's supposed independence in the AI race.

Station F’s French Ambitions vs. the Global AI Funding Race

Station F—once a cavernous rail depot—has morphed into the beating heart of France’s tech scene, complete with rooms piled high with Lego bricks, an arcade, and a swanky Italian restaurant serving up famous pizza. The recent visit from OpenAI’s Sam Altman and French President Macron during the Paris AI Summit served as a reminder of just how big a draw this place is for founders, investors, and Big Tech alike.

Station F admits 50 new start-ups every month, offers mentorship hubs from tech giants like Apple and Google, and collectively raises over €1B a year—a figure equal to about 15 percent of all start-up funding in France. Believe it or not, that’s also more than the entire Italian tech scene secured the previous year (which might say more about Italy’s culinary priorities than its tech ambitions).

AI stands out as the star attraction among the startups there. In fact, 34 of its top 40 ventures have AI at their core.

While it’s great to hear that France is growing its AI development scene, it’s still far behind the leading countries. The United States has consistently led global AI venture capital funding, followed by China, then the UK.