NewsNews

Technology news

  1. 1

    US bans DeepSeek AI from civilian government agencies

    The US Department of Homeland Security has banned use of DeepSeek's AI within civilian government agencies, including the Treasury, State and Justice departments, as well as any technology built with its models. The order was made public on Monday but took effect at the end of July, and agencies within the Department of Defense are excluded. The DHS notice did not detail the security concerns behind the move. The ban follows criticism of Chinese AI from the Trump administration and US firms. White House tech lead Michael Kratsios wrote in an internal memo of "malicious exploitation" of American AI firms, mainly by Chinese rivals, while Anthropic and OpenAI have claimed some Chinese firms, including DeepSeek, copied their models. Meanwhile, US companies including Snowflake, Vercel and OpenRouter incorporate DeepSeek into their offerings.

    Key facts

    • DeepSeek said its debut model cost about $6m (£4.5m) to train, versus roughly $100m for OpenAI's GPT-4.
    • The most recent DeepSeek version was released last week, and users have praised it online.
    • DeepSeek's free chatbot is among the most downloaded apps in the Apple App Store rankings.
    • Anthropic and OpenAI are both poised to list on the public stock market in the coming months.
    • The DHS, DeepSeek and the Chinese embassy in Washington were contacted for comment.

    Commentary

    The ban is narrower than its scope suggests: it covers civilian federal offices, while US companies still build DeepSeek into products sold to their customers. With no security rationale published, the restriction rests on accusations from the White House and rival firms rather than disclosed evidence, which makes the Department of Defense exemption the detail to watch.

    Sources: BBC

  2. 2

    China's third-quarter growth picks up as AI-driven exports offset weak domestic demand

    China's economy grew 4.5% in the third quarter, up from the previous period but below the 5% expansion recorded from January to March, official Chinese data showed. The pickup followed a three-year low and was driven by exports, as global demand for chips and components used in AI data centres eased pressure on Beijing. Weak spending at home persists, with consumers pessimistic and unwilling to spend, according to Oxford Economics researcher Sheana Yue. Moody's Analytics economist Dave Chia said the effects of Beijing's targeted business support have yet to be seen and flagged reliance on a single external engine as a risk. Trade tensions with the US and European Union, and doubts over whether AI investment will deliver promised profits, cloud the outlook.

    Key facts

    • Trade surplus is on track to overtake last year's record $1.2tn (£900bn)
    • Sales of high-tech goods rose 14.3% in the first nine months of the year
    • Battery and industrial robot production each grew by more than 29%
    • Property investment fell by more than 20% in the first nine months
    • Beijing's annual growth target is 4.5%-5%, with the latest figure at the bottom end

    Commentary

    The rebound rests on one engine, AI-linked exports, not on a recovery in household spending or property. That leaves Beijing's stimulus, whose effects are not yet visible, carrying more weight if overseas demand or AI spending falters, and it puts the cost on manufacturers already facing Western pushback against cheap Chinese goods.

    Sources: BBC

  3. 3

    AI stocks lead Wall Street rebound as Asian tech shares fall

    AI stocks led a Wall Street rebound on Monday, recovering some of last week's steep losses, which followed worries about the rising cost of borrowing money. Nvidia, the largest U.S. stock by market value, was among the strongest forces lifting the market, while most stocks outside AI and tech fell. Much of the AI spending on chips and data centres has been financed with borrowed money, and Treasury yields have hit multi-year highs. In Asia, Japanese and Chinese shares slipped on Tuesday as AI spending worries persisted, while South Korea's Kospi rose on chip rebounds.

    Key facts

    • S&P 500 rose 0.4% to 7,690.73 on Monday, coming off its worst week since June.
    • Nasdaq composite rose 1.1%, while the Dow Jones Industrial Average fell 36.38 points to 51,966.03.
    • 10-year Treasury yield rose to 5.14%, after topping 5.30% earlier this month.
    • Nikkei 225 fell 1% to around 64,300 on Tuesday, approaching one-month lows.
    • Associated Press put Kospi's drop last week at 14.5%; BullionVault said more than 11%.

    Commentary

    The rebound is narrower than the headline suggests: a handful of AI names lifted indexes while most other stocks fell, leaving the market dependent on one sector. The real test is Meta Platforms and Lam Research reporting Wednesday, because the debt-funded spending behind AI is exactly what higher borrowing costs threaten.

    Sources: BNN Bloomberg, BullionVault, Yahoo! Finance Canada, TradingView

  4. 4

    Chip stocks including Micron rise in morning trading as analysts raise price targets

    Shares of chipmakers including Micron, Applied Materials, Lam Research and Marvell Technology jumped in the morning session as investors returned to stocks that had been beaten down. The rebound followed a 4.5% fall in the Nasdaq the previous week, which ended a four-week winning streak, according to CNBC. That slide was attributed to doubts over the durability of the artificial intelligence build-out after Anthropic reported third-quarter revenue of $16.8 billion, alongside widening credit spreads at debt-funded compute providers. Sentiment improved after JPMorgan upgraded Intel to neutral from underweight, and the 10-year Treasury yield retreated. Separately, D.A. Davidson raised its Micron price target to USD 3,000 from USD 2,100 on October 7 while keeping a Buy rating, as CNBC reported.

    Key facts

    • Lam Research shares are 36.1% below their 52-week high of $433.33 from June 2026.
    • Micron reported fiscal fourth-quarter revenue of USD 54.23 billion on September 30, 2026, up from USD 11.32 billion.
    • Micron guided fiscal first-quarter 2027 revenue to USD 61.5 billion, plus or minus USD 1.5 billion.
    • Micron traded at EUR 800.75 on Lang & Schwarz on October 19, 2026, 0.03% below the Xetra close.
    • Lam Research has had 52 moves greater than 5% over the last year.

    Commentary

    The rebound rests on lower bond yields and one broker upgrade, not on answers to the doubts about debt-funded data center spending that triggered the drop. Analysts are lifting Micron targets on valuation after management meetings, so the rally depends on whether next quarter's guidance is delivered.

    Sources: TradingView, ad-hoc-news.de, Stock Titan

  5. 5

    Anthropic and Nscale Discuss a Second Cloud Computing Deal at Monarch Campus

    Anthropic and Nscale are reportedly discussing a second cloud-computing agreement worth about $45 billion, according to Bloomberg, which published the account on October 19. The proposed deal would cover another building at Nscale's Monarch campus in West Virginia. Reuters reported the Bloomberg account but said it could not independently verify the talks, and neither company had responded to its requests for comment. Neither the price nor the capacity has been confirmed. The talks are separate from an arrangement of roughly the same value reported in August, and nothing has been signed. Nscale describes Monarch as a development that is planned rather than fully online, so the proposed building depends on a campus still being built.

    Key facts

    • The proposed second Monarch building would use roughly 460 megawatts of capacity with Nvidia's forthcoming Vera Rubin chips.
    • Anthropic would have a right of first refusal for a third building, Reuters said, citing Bloomberg's sources.
    • Nscale says Monarch's first phase is designed for 1.35 gigawatts, with delivery targeted by early 2028.
    • Akamai disclosed an $11.6 billion CPU-focused agreement with Anthropic in September.
    • Nscale says Monarch will generate power on-site and use closed-loop cooling.

    Commentary

    The new element is less the price than the location: a second building at Monarch would tie Anthropic's capacity to one campus's construction schedule. That makes delivery timing, not the headline value, the risk a customer carries if the talks become a contract.

    Sources: TechEngage

Today's video

Also posted to: