Good morning. Okay, we’re doing it. We are diving into a bit of quantum computing on this fine Thursday.

If you want to give yourself a bit of a refresher on what quantum even is, check out this recent chat I had with Dr. Jerry Chow, IBM’s director of quantum infrastructure.

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

• 🌎 AI for Good: Plastic management 

• 👁️‍🗨️ Microsoft says it made a huge quantum breakthrough

• 📊 Humane’s devolution and a retreat from ‘revolution’

AI for Good: Plastic management 

There are a lot of different types of plastic, a reality that makes plastic recycling both time-consuming and expensive. See, the plastic needs to get sorted first, something that requires both people and time. 

What happened: The National Institute of Standards and Technology has been developing a plastic-sorting method that combines machine learning with lasers to improve efficiency. 

• The researchers have been exploring near-visible infrared light (NIR), a technology that can scan plastics and rapidly define its type. 

• The team leveraged machine learning to identify these NIR “fingerprints,” enabling a computer to identify the material in a given piece of plastic, and sort it accordingly. 

Why it matters: Though it’s still early days, the promise, according to NISt, is simple: “If we can sort those effectively, we can reuse them with fewer processing problems, and recycling will become much more profitable. Then, hopefully, the profits can drive even better recycling habits, and we can start bending our linear economy into a circular one.”

Certified e-Bikes, Unbeatable Prices—Up to 60% Off!

Shop top-tier e-Bikes from brands like Aventon, Specialized, and Cannondale—certified, warrantied, and up to 60% off retail.

Upway makes it easy to buy or sell your e-Bike with fast delivery and expert support.

Plus, take $150 off your next ride with code thedeepview.

Upgrade your ride today. Shop now.

Microsoft says it made a huge quantum breakthrough

The news: Microsoft on Wednesday announced Majorana 1, its first quantum computing chip. According to Microsoft, the chip, leveraging a new type of material called a topoconductor, will allow Microsoft to build a “truly meaningful” quantum computer in years, not decades. 

This approach of Majorana-based topological qubits is one that Microsoft has been pursuing for nearly 20 years. 

(Apologies in advance for all the physics-y jargon. Here we go). 

A quick note on quantum: Unlike classical computers, quantum computers use quantum processors and qubits, rather than silicon processors and bits. Simply, since quantum computer are based on the physics of quantum mechanics, researchers believe that quantum computers will be able to conduct calculations that classical computers can’t. 

• The theoretical idea is that topological qubits would be more reliable than traditional qubits, due to the arrangement of topological particles (picture a braided rope). And, on top of that, Microsoft said that the architecture that enabled its new chip “offers a clear path to fit a million qubits on a single chip that can fit in the palm of one’s hand.” 

• Microsoft said that that scale is the base threshold of what is required for quantum computing to become actually impactful: “all the world’s current computers operating together can’t do what a one-million-qubit quantum computer will be able to do.” 

So far, Microsoft has placed eight topological qubits on a chip designed to hold one million. 

Microsoft is far from the only tech giant playing in quantum. Google recently announced its own breakthrough quantum chip, and IBM’s been pushing advances in quantum for years, now; its latest quantum processor can fit 156 qubits.

Here’s where it gets complicated: Dr. Michael Jarret, the associate director of George Mason University’s Quantum Science and Engineering Center, told me that the Majorana particles Microsoft has been chasing “are inherently robust against errors, which is a major bottleneck in scaling existing quantum technology.” 

• “The theory community largely believes Majorana particles exist, but finding them has been a long-standing theoretical challenge,” he said. “Microsoft appears to have some evidence that they can perform measurements necessary for and consistent with Majorana-based quantum computing, however, whether the particles have actually been discovered in their experimental approach still appears to be an outstanding question (even after this paper).”

• It’s an impression echoed in the peer review for the paper associated with the advance, in which one referee wrote that “this work rests on an extremely fragile ground,” with another writing that “the relationship to Majorana physics is not completely certain and needs some serious scrutiny.” 

Jarret said that it seems like a “major advancement in Microsoft's quantum agenda,” though added that he’s left with a “lot of questions.” 

It’s a possibly promising step deeper into quantum; it’s not a solution that itself, today, makes quantum work at scale. 

Still, shares of Microsoft lifted more than 1% on Wednesday, amid affirmations from Jason Zander, a Microsoft executive vice president, that when the legitimate quantum computers arrive, we’ll be able to “invent some new molecule, invent some new drug.”

• An Elon Musk staffer developed a chatbot designed to help DOGE make government ‘less dumb,’ according to TechCrunch. It’s powered by xAI.

• Meta is building a 31,000-mile-long undersea cable that, according to the company, will ensure that the “benefits of AI” are accessible around the globe. It’s a reminder of the significance of the hardware necessary for the software that sits at our fingertips.

• Apple unveils cheaper iPhone 16e powerful enough to run AI (CNBC).

• Microsoft’s Xbox AI era starts with a model that can generate gameplay (The Verge).

• Not even DOGE employees know who is legally running DOGE (Wired).

• Meta’s brain-to-text tech is here. We are not remotely ready (Vox).

• Trump and Zelensky Escalate War of Words Amid Debate Over Peace Talks (WSJ).

Humane’s devolution and a retreat from ‘revolution’

In 2023, Humane’s AI Pin was featured in Time Magazine’s roundup of the best inventions of the year. The promise — even before it was officially released — was “a screen-free future.” 

Then, in 2024, the Pin — described as a “true game-changer” — won RedDot’s award for Best Innovative Product. The future, they said, was here. 

But it isn’t anymore. 

On Tuesday, HP acquired Humane for $116 million — the legacy tech firm is taking Humane’s talent, patents and “key AI capabilities” and merging them into a new unit called HP IQ, an AI innovation lab. 

The AI Pin itself didn’t make the cut. 

Humane said in a separate note that, “effective immediately, we are winding down the consumer AI Pin as our business priorities have shifted.” The Pin is no longer for sale; Pins that have already been purchased will be disconnected from Humane’s servers on Feb. 28, the same day that all customer data will be deleted. 

The company will only offer refunds on Pins that are still within their 90-day return window from their original shipment date. So, get ready for an influx of high-tech e-waste.

None of this is really that surprising. 

• The Verge reported in August that, shortly after its release, “more AI pins were returned than purchased,” a reality that is pretty much in line with the reviews the device was getting. Marques Brownlee called it the “worst product” he’s ever reviewed, saying: “there’s so many things bad about it. It’s so bad that it’s kind of distracting to understand what the point of the device is.”

• Humane had raised more than $200 million from folks including Marc Benioff (the owner of Time and CEO of Salesforce) and OpenAI’s Sam Altman. 

“I think the main lesson is that distribution matters a lot,” Bain Capital Ventures partner Rak Garg told me. “Humane tapped into what could be possible, but it was mired by execution challenges (slow, buggy, etc etc). Customers, especially consumers, have very little tolerance for AI products that aren’t able to live up to the marketing reel.”

With this all in mind, let’s look back at a different side of 2023. Fintech firm Klarna halted its hiring in late 2023, letting its workforce shrink by about 20%. See, Klarna’s CEO — Sebastian Siemiatkowski — figured that generative AI chatbots were good enough to replace human customer service agents, so the company replaced around 700 of them with GenAI. 

Siemiatkowski said a few months ago that “AI can already do all of the jobs that we, as humans, do. It's just a question about how we apply it and use it.”

But it seems that, on the ground, that reality has not played out.

• “We just had an epiphany: in a world of AI nothing will be as valuable as humans,” Siemiatkowski wrote last week. “Ok you can laugh at us for realizing it so late, but we are going to kick off work to allow Klarna to become the best at offering a human to speak to!!!”

• “We will continue to invest and improve our AI support,” he added. “But the truth is the cost savings it generates and improvement in quality allows us to double down on making sure the human service part of Klarna becomes even better.”

So, AI didn’t kill customer service, after all. 

And AI didn’t kill the iPhone, either. 

As predicted, I would say that this marks the beginning of the bubble burst. I think we’ll see a lot more of this. 

From the very beginning, the challenge with the business of AI was twofold: one, massive capital requirements, and two, technical limitations that make efficacy and usability hazy at best. 

As Garg told me last year, “the way we think about the bubble is, is there a market big enough to sustain whatever valuation that we invest at?” 

We’ve got companies raising enormous amounts of money at enormous valuations, despite the fact that the actual market for these products likely can’t sustain companies of that size, despite the fact that, two years and hundreds of billions of dollars into this race, we don’t have any sort of significant returns. 

There was so much hype around Humane’s AI Pin. But the market couldn’t sustain it, due to some combination of poor efficacy, little interest and/or high costs. This theme of tremendous costs and challenging implementation is one that repeats up and down the AI ecosystem; the major developers are not immune to it. 

I think the AI bubble will be slowly bursting for a long time, and on the other side of it, I think we’ll see a calmer iteration of generative AI that gets implemented into software layers in targeted ways. I think in many cases, it simply cannot stand on its own.

🤔 Your thought process:

Selected Image 1 (Left):

• “The AI image’s farm crops are too 'perfect' in the rows they are planted, the real image has real crops with imperfections. Also, the lighting in the AI image is also just too 'perfect', it's very hard to get all of that perfectness in one shot.”

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

Here’s your view on Grok 3:

33% of you think it’s about even with the SOTA.

17% think it’s the best LLM out there and 15% think it sucks.

And 20% (like me) are so, so bored of chatbots.

Seems even:

• “It reminds me of the ‘deep learning’ craze when all the hype and focus was moving a needle on a single handwriting benchmark from 87.3% to 87.5% just to be state-of-the-art. I'm looking forward to another disruptor to give a new focus to research.”

So bored of chatbots:

• “Yep, still waiting on the killer app. The Lotus 123 of LLMs. ( Yes I'm that old.)”

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.