In the indubitably human-created words of Ron Burgundy: That escalated fast.

Yesterday saw a huge selloff in US tech stocks after a previously obscure Chinese AI chatbot app, called DeepSeek, shot to the top of the Apple App Store. Why the fuss? Well, DeepSeek managed to show that it had created a high quality chatbot but with far fewer resources than US artificial intelligence titans like OpenAI — and at a much cheaper price point. So suddenly the US companies looked like bloated, lumbering dinosaurs being out-maneuvered by an agile little Chinese company, their competitive moat a mere puddle. 

But what does DeepSeek’s sudden rise to fame really tell us about the AI market?

A Little Friendly Competition

First off, the US bid to freeze China out of the AI big leagues has experienced a rude awakening. While DeepSeek’s popularity rush gave investors an unpleasant surprise, Dr Keegan McBride, an expert in AI policy at the Oxford Internet Institute, told The Daily Upside that serious AI competition from China was overdue: “There are a lot of people who got caught off guard, but part of the reason for that is because the United States had convinced itself that closed-source models were far ahead of open-source ones, that there are these huge compute barriers, these export controls in place, and they made the mistake of thinking that this was a permanent fix for curtailing Chinese advancements, rather than trying to slow it down.”

That said, there is a degree of hype around DeepSeek and it’s not the end of American AI tech as we know it. The numbers that it gave out, which spooked tech investors, come with some big asterisks:

• DeepSeek said it only spent $5.5 million “training” the model that has performed so well, but that’s stripping out pretty meaningful overhead, like staff salaries. 
• It also trained its model on a stockpile of Nvidia chips, which have since had export controls imposed on them, so arguably the US effort to box out Chinese competition hasn’t really started to bite yet.

On the other other hand, the technological capability demonstrated by DeepSeek is not to be sniffed at. “Is it cool? Yes. Is it important? Yes. Is it sort of like regime-shattering and an everything-is-burning-down moment? No, not at all,” McBride said. He added that although sparing compute power is part of what’s making DeepSeek a star now, it will become an issue if and when the company begins to scale. Addressing the stock rout, McBride said American AI stocks were likely overvalued to begin with. 

Pricing In Price Wars: Joanna Bryson, professor of ethics and technology at the Hertie School in Berlin, told The Daily Upside that DeepSeek’s impact could be to change the narrative around how AI companies compete. While companies previously were vying to capture resources and infrastructure at huge expense, DeepSeek has flipped the script. “A price war is the best thing that could happen to the industry, honestly,” Bryson told The Daily Upside. “It’s a way to cut through and say: ‘No, stop consuming as many resources as possible.’” Her words were echoed in energy stocks, which took a dive on Monday as markets grappled with the idea that tech companies may stop chasing after power supply with the same verve.

Behind the blinding white light of the trillion-dollar artificial intelligence wipeout that was the talk of markets Monday was a spot of unabashedly good AI news.

Manas AI, an oncology startup co-led by LinkedIn founder Reid Hoffman, raised $24.6 million to use the technology to identify drug candidates to help treat breast cancer, lymphoma and prostate cancer.

Bet on Yourself

The promise of generative AI in medical research — that it should be able to parse mass troves of biological data to pinpoint potential new treatments — is straightforward. But the results, so far at least, leave a lot to be desired. No drug candidates have been approved from this method, which other startups are also employing. AI has found use in surgeries, medical imaging and in identifying protein structures, which is key to drug research.

Hoffman’s co-founder, Columbia University professor and cancer researcher Siddhartha Mukherjee told The Wall Street Journal that he’s confident in AI’s ability to model treatments, and at a pace beyond what humans are capable of: “AI was a mechanism to scale — vastly — the kinds of work that we were doing in any one laboratory, and across diseases.” Keeping the faith is one thing, the other is money:

• Manas’ funding round was led by Hoffman himself. Joining him were venture capital giants General Catalyst — which has poured hundreds of millions into healthcare and plans to invest much more — and Greylock Partners — which turns 60 this year, making it an elder statesman of the VC world.
• While AI research into drug discovery has come up short on transforming the industry, investors are still placing bets on its future. Startups in the sector raised $3.3 billion last year, five times their 2023 haul, according to PitchBook. Of note were a $130 million round for nature-based drug discovery startup Enveda and $300 million for quantum AI startup SandboxAQ, which is partly focused on drug discovery.

Know Your Worth: In unrelated biotech news, Sage Therapeutics rejected a $442 million buyout offer from its longtime pharma giant partner Biogen on Monday. The Cambridge, Massachusetts, firm’s shares declined 75% in value last year after a series of candidates, including potential treatments for Huntington’s and essential tremor, flopped at trial. While Sage said it’s open to a deal, it thinks it’s worth more than that as it doubles down on upping the adoption of its approved postpartum depression drug Zurzuvae.

Say this much for a housing market that’s been slower than a turtle on vacation: It’s making way for the new at least. 

A report from the US Commerce Department published Monday showed that sales of new homes increased more than expected in December (and were higher than previously thought in November). But sales of old homes remain historically low — keeping the overall housing market as slow as molasses in January.

The Price is Not Right

Here’s the good news: Sales of new single-family homes rose 6.7% year-over-year in December, to a seasonally adjusted rate of 698,000 homes, per the Commerce Department. Meanwhile, November’s annual rate was revised upward to 674,000 units, a 10,000-home increase. The total number of new home sales for all of 2024, meanwhile, climbed to an estimated 683,000, up 2.5% from 2023 and the highest mark in three years.

But that’s just new home sales, which account for about 14% of total home sales in the US. Look under the hood, and the housing market still looks downright stagnant:

• In data published Friday, the National Realtors Association said that just 4.06 million existing homes were sold in the US in 2024, down 0.7% year-over-year and good for the lowest mark since 1995. Existing home sales in December were about 20% lower than the average monthly sales pace in 2019. 
• Blame historically low supply and strong demand that’s keeping prices comparatively high. The national median existing home price in December was just over $404,000 — down from June’s all-time $426,900 record, but still 6% higher than December 2023. 

Inventory Check: Supply might be catching up to demand. Key word: might. Per the Commerce Department, the inventory of new homes reached 494,000 units in December, up around 6,000 from November and good for the largest pool of empty homes since December 2007. Meanwhile, December also saw the completion of construction of 118,000 new homes, the most since August 2009. But builders are also waiting to break ground on some 108,000 homes, an all-time high. As Stephen Stanley, chief US economist at Santander US Capital Markets, tells Reuters, that just means builders are likely to “remain in cautious mode” to thin out supply.