This Week in AI: Microsoft Sees Quantum Breakthrough as Ai Pin Deflates

If you’re interested in artificial intelligence (AI) news, here are the the latest in AI updates.

Let’s start with Microsoft’s quantum moment. It has created the world’s first ‘topoconductor,’ which is a new type of matter — not solid, liquid or gas — that has only previously existed in theory. The topoconductor forms the core of Majorana 1, its new quantum chip.

“We believe this breakthrough will allow us to create a truly meaningful quantum computer not in decades, as some have predicted, but in years,” Microsoft CEO Satya Nadella wrote in a Wednesday (Feb. 19) LinkedIn blog post.

Quantum computing has been in development for decades, but its exponentially greater computing power has not found practical use due to instability, errors and cost. It also needs to scale to be useful.

Microsoft said its topoconductor, or topological superconductor, leads to a more stable qubit, or quantum bits used for processing information in quantum computers. This is opposed to today’s “classical” computer bits of 1s and 0s that are manipulated to perform computations.

But qubits aren’t stable. When they are destabilized, they turn into an ordinary bits and lose their quantum abilities that enable them to perform exponentially greater than classical computers.

“The qubits created with topoconductors are faster, more reliable and smaller,” Nadella said. “They are 1/100th of a millimeter, meaning we now have a clear path to a million-qubit processor” or chip.

Nadella added that its palm-sized quantum chip can solve problems that “all the computers on Earth today combined could not!”

In December, Google unveiled its new Willow chip that can “reduce errors exponentially” as it scales using more qubits, according to a company blog post. It was able to do a benchmark calculation in less than five minutes that would take today’s fastest supercomputer 10^25 (septillion) years to do.

AI Startup Figure Debuts New Robots

Figure’s humanoid robots are showing off skills we’ve always wanted robots to have, such as putting away our groceries, among other ordinary tasks.

The Nvidia-backed robotics AI startup unveiled Helix on Thursday (Feb. 20), a generalist vision-language-action model that gives a robot the combined ability to see, understand language and exercise control.

Figure said its robots can now pick up anything, even small household objects that they have never encountered before.

The video released by Figure shows its robots understanding the homeowner’s voice command to put away groceries in their right place. They pick up the items and decide where to place them — some in the fridge and others in cabinet drawers.

This is a big deal because the hardest part of making humanoid robots useful isn’t just getting them to move like humans — it’s getting them to generalize and adapt to new tasks in dynamic environments.

Figure aims to get its robots into factories, warehouses and ultimately, homes. If they pull this off at scale, we could be looking at a future where C-3PO-type robots of Star Wars fame become more than just science fiction.

Musk’s xAI Introduces Grok 3

Meanwhile, Elon Musk’s xAI on Wednesday launched Grok 3, its latest attempt to go toe-to-toe with OpenAI, Anthropic, Google and other AI rivals.

While early versions of Grok felt more similar to ChatGPT, Grok 3 outperformed top-of-the-line AI models from OpenAI, Google, Anthropic and new Chinese startup DeepSeek.

Grok 3 was trained on double the computing power as Grok 2: a cluster of 200,000 Nvidia H100 GPUs running on its own Colossus supercomputer.

The real question is how long Grok can hold on to its lead in performance. Meta is set to launch Llama 4, and OpenAI is releasing GPT-4.5 and GPT-5. Anthropic is also supposedly coming out with an updated model soon.

Ai Pin-Maker Humane Bought out by HP

Finally, let’s talk about Humane’s Ai Pin, a wearable AI-powered smart device that its creators hoped would replace smartphones. It was a square device that users attached to the shirt like a brooch but using magnets. The idea was to evoke the Star Trek communicator.

The AI Pin projected information onto your hand and could send messages, answer calls, search the web, play music, take photos, translate conversations and more. Time magazine even named it among the best inventions of 2023.

The $699 device — later discounted to $499 — launched less than a year ago but immediately faced negative reviews from early adopters: The AI was slow. The interface was clunky. The device overheated. And don’t forget the $24 a month subscription fee (excluding taxes and other fees) just to use it.

Humane had some serious hype going in — a founding team of ex-Apple veterans and funding from OpenAI CEO Sam Altman, SoftBank, Microsoft and Salesforce CEO Marc Benioff (who owns Time).

But none of that mattered once people actually tried using it. The AI Pin was impractical and expensive.

However, HP is buying the startup for $116 million, a purchase that includes its technology, technical staff and patents, both granted and pending. The Humane team would become HP’s new AI innovation lab called HP IQ, tasked with making its computers, printers and scanners smarter.

In the meantime, the 10,000 early adopters of the AI Pin were told their devices will no longer work after Feb. 28.

The post This Week in AI: Microsoft Sees Quantum Breakthrough as Ai Pin Deflates appeared first on PYMNTS.com.