Nvidia shares fell below the critical $200 threshold, closing at $199.48 amid reports that Google is preparing new in-house AI inference chips. Despite the 0.79% daily decline, the stock maintains strong technical positioning above all key moving averages clustered around $181-$183, with MACD still flashing a buy signal and ADX indicating a continuing upward trend despite current volatility.

Google takes aim at Nvidia’s fastest-growing market

While Nvidia’s stock treads water, Google is making its most direct push yet into the chip market.

The Alphabet-owned company is preparing to announce a new generation of tensor processing units, known as TPUs, at its Google Cloud Next conference in Las Vegas this week, with a focus on inference: the process of running AI models once they have already been trained.

“It now becomes sensible to specialize chips more for training or more for inference workloads,” Google Chief Scientist Jeff Dean said. The company is “looking at a whole bunch of different things,” he added, including how fast it can deliver AI results to users.

Big names are now in for the TPUs. Anthropic has signed a contract for 1 million TPU’s while Meta is using them through Google’s cloud as part of a multi-billion-dollar agreement. In the upcoming Google conference, Citadel Securities will be talking about how TPU’s are faster at training models than GPUs. This isn’t where it ends. Abu Dhabi’s G42 is also in discussion to access them.

Google is also loosening the rules around TPU access, letting some customers run the chips inside their own data centers and supporting outside tools like PyTorch, rather than locking users into Google’s own software stack.

Not to forget, OpenAI is also growing frustrated by Nvidia’s inference hardware and looking for alternatives, as reported by Cryptopolitan previously.

Billions are flowing into chip startups

Google is not the only one sensing an opening. AI chip startups pulled in $8.3 billion globally in 2026, according to data from Dealroom, putting the sector on track for a record year. In the U.S., Cerebras raised $1 billion in February. MatX, Ayar Labs, and Etched each secured $500 million rounds. In Europe, Axelera and Olix both raised over $200 million.

“It’s no longer a niche bet,” said Carlos Espinal of European VC firm Seedcamp. “It’s becoming a core part of how people think about AI infrastructure.”

Samsung-backed South Korean startup Rebellions raised $400 million at a $2.34 billion valuation, led by Mirae Asset Financial Group and South Korea’s state-backed National Growth Fund. The company has raised $650 million in the past six months alone, more than 75% of its total funding, and is now targeting U.S. customers and preparing for a public listing.

Its Rebel100 chip is built specifically for inference.

One constraint is memory. High-bandwidth memory is tight across the industry, and prices have risen sharply. “Memory is not very easy to get. But our demand is so huge,” Park said. Rebellions has an edge there: both Samsung and SK Hynix are investors, giving it better access than most rivals.

The crypto card with no spending limits. Get 3% cashback and instant mobile payments. Claim your Ether.fi card.