"Not even AI can predict the future (yet), nevertheless two of our tech editors have taken a look ahead to what they think will be big in 2025.
As 2022 drew to a close the outlook was bleak for the cryptocurrency business.
The scandal rattled confidence in the whole sector.
It seemed that cryptocurrencies would remain a niche product, with an enthusiastic but relatively limited following.
But just a few months later and the industry was fizzing with optimism again. Behind the enthusiasm - the success of Donald Trump in the 5 November presidential election.
The feeling was that he would be more favourable to the cryptocurrency sector and, so far, that seems to be the case.
Mr Atkins is seen as being far more pro-cryptocurrency than outgoing head, Gary Gensler.
That announcement helped the value of one bitcoin, the biggest of the numerous cryptocurrencies, surge through $100,000.
"With Trump winning you can imagine in 2025 you'll get proactive regulation. You'll get removal of some negative regulation, which will then allow banks and other institutions into the space," says Geoffrey Kendrick, global head of digital assets research at Standard Chartered.
As AI tools move into our phones – Apple, Google, and Samsung have all launched services that can edit photos, translate languages, and carry out web searches – we are at the start of an era in which AI becomes an intrinsic part of our digital lives and increasingly helpful on a personal level.
That’s if we allow it, because it does require a bit of a leap of faith.
Let’s take diary management as an example. An AI tool efficiently can manage your diary for you if you allow it to access it. But how far should this go?
In order to be truly useful, does that mean it also needs to know who you would rather avoid meeting, or relationships you want to keep secret, and from whom?
Do you want it to provide you with summaries of counseling sessions, or medical appointments?
It’s deeply personal information, and potentially both hugely embarrassing and extremely valuable if some glitch meant it was shared. Do you trust the big tech firms with that kind of data?
Microsoft is pushing hard at this particular door. It got into trouble in 2024 for demoing a tool called Recall, which took snapshots of laptop desktops every few seconds, in order to help users locate content they’d seen but couldn’t remember where.
It has now made a number of changes to the product – which was never launched – but stands by it.
“I think we’re moving to a fundamentally new age where there will be ever present, persistent, very capable co-pilot companions in your everyday life,” the firm's head of AI, Mustafa Suleyman told me recently.
Despite the challenges, Ben Wood, chief analyst at technology research company CCS Insight, expects that more personalized AI services will emerge in 2025.
"The output will be continuously updated by drawing on evolving data sources, such as emails, messages, documents, and social media interactions.
“This will allow the AI service to be tuned specifically to a person's communication style, needs, and preferences," he says.
But Mr. Wood accepts that letting AI loose on your personal information will be a big step.
"Trust will be essential," says Mr. Wood.
The more money pours into AI, then the more data centers will need building.
Training and running AI requires a lot of computing power, and works best with the latest computer chips and servers.
In Europe alone, between 2024 and 2028, data center capacity is expected to grow by an average of 9% annually, according to property services company Savills.
But those new facilities are unlikely to be built in the current data center hubs like London, Frankfurt, and Amsterdam.
In the UK, cities like Cambridge, Manchester, and Birmingham could well be home to the next wave of data center construction.
Elsewhere, Prague, Genoa, Munich, Dusseldorf, and Milan are likely to be considered in Europe.
At the heart of some of those new data centers will be the latest computer chip from Nvidia, the company that dominates the market for chips used for AI.
The new chip should allow tech firms to train AI four times faster and see AI operate 30 times faster than current computer chips, according to Vivek Arya, senior semiconductors analyst at Bank of America Securities.
But other customers might struggle to get their hands on the super chip, with "supply constrained in 2025," according to Mr. Arya.