Page 4 of 113

Alibaba introduces Qwen 2.5-Max AI model, claims edge over DeepSeek

January 30, 2025 /

China’s Alibaba Group has launched an upgraded version of its Qwen 2.5 AI model, claiming it outperforms models from DeepSeek, OpenAI, and Meta, as competition in the AI market intensifies.

“Qwen 2.5-Max outperforms … almost across the board GPT-4o, DeepSeek-V3 and Llama-3.1-405B,” Alibaba’s cloud unit said on its WeChat account, according to Reuters.

On its GitHub page, the company showed benchmarking results indicating that its instruct models – designed for tasks like chat and coding – mostly outperformed GPT-4o, DeepSeek-V3, and Llama-3.1-405B, while performing comparably to Claude 3.5-Sonnet.

The launch follows DeepSeek’s disruptive entry into the market, marked by the Jan 10 debut of its AI assistant powered by the DeepSeek-V3 model and the Jan 20 release of its open-source R1 model.

The Chinese startup’s low-cost strategy has shaken Silicon Valley, sending tech stocks lower and prompting investors to question the sustainability of major US AI firms’ high-spending approach.

China’s AI race heats up

Alibaba’s launch coincided with the Lunar New Year holiday, a time when much of China is on break, underscoring the growing competitive pressure from DeepSeek.

DeepSeek’s rapid ascent over the past three weeks has intensified rivalry not only with global players but also among Chinese tech firms.

“The AI model war is no longer just China versus the US – competition within China is also intensifying as companies like DeepSeek, Alibaba, and others innovate and optimize their models to serve a high-scale domestic market,” said Neil Shah, partner and co-founder at Counterpoint Research. “Chinese companies are being pushed to innovate further due to resource constraints, including limited access to the most advanced semiconductors, global-scale data, tools, infrastructure, and audiences.”

The race for frugal AI

The race to develop high-performance, cost-efficient AI models is intensifying, challenging the business strategies and pricing structures of major US hyperscalers and AI firms as they seek to recover billions in investment.

“This gives enterprise buyers and decision-makers more leverage, increasing pricing pressure on AI applications built with more expensive underlying models,” Shah said. “Such breakthroughs will force enterprises to reconsider, or at least rethink, the economics of AI investments and their choice of models and vendors.”

DeepSeek is driving immediate pricing considerations in two key areas of AI – raw token costs and model development expenses. These factors may force AI companies worldwide to consider optimizing their models to remain competitive. 

“DeepSeek’s success also highlights the power of open source, strengthening the argument that open-source AI could become a dominant market later,” said Hyoun Park, CEO and chief analyst at Amalgam Insights. “If that happens, companies with strong open-source business models for enterprises – such as IBM Red Hat and Canonical – could step in and rapidly scale AI-related managed services.”

The geopolitics advantage

Geopolitics remains a wild card for Western AI firms, potentially tilting the market in their favor by restricting the adoption of Chinese models in certain regions.

At the same time, China is likely to tighten controls on the use of Western AI models, mirroring restrictions seen with other tech applications.

Enterprises will first assess whether these models comply with global privacy and regulatory standards before adopting them at scale, said Sharath Srinivasamurthy, associate vice president of Research at IDC.

“DeepSeek’s advancements could lead to more accessible and affordable AI solutions, but they also require careful consideration of strategic, competitive, quality, and security factors,” Srinivasamurthy said.

However, China’s substantial investment in AI research and development is only beginning to yield results, according to Srinivasamurthy. Other Chinese firms, like Alibaba, which have also been investing in AI in recent years, may soon start launching their own models.

Alibaba introduces Qwen 2.5-Max AI model, claims edge over DeepSeek

January 30, 2025 /

China’s Alibaba Group has launched an upgraded version of its Qwen 2.5 AI model, claiming it outperforms models from DeepSeek, OpenAI, and Meta, as competition in the AI market intensifies.

“Qwen 2.5-Max outperforms … almost across the board GPT-4o, DeepSeek-V3 and Llama-3.1-405B,” Alibaba’s cloud unit said on its WeChat account, according to Reuters.

On its GitHub page, the company showed benchmarking results indicating that its instruct models – designed for tasks like chat and coding – mostly outperformed GPT-4o, DeepSeek-V3, and Llama-3.1-405B, while performing comparably to Claude 3.5-Sonnet.

The launch follows DeepSeek’s disruptive entry into the market, marked by the Jan 10 debut of its AI assistant powered by the DeepSeek-V3 model and the Jan 20 release of its open-source R1 model.

The Chinese startup’s low-cost strategy has shaken Silicon Valley, sending tech stocks lower and prompting investors to question the sustainability of major US AI firms’ high-spending approach.

China’s AI race heats up

Alibaba’s launch coincided with the Lunar New Year holiday, a time when much of China is on break, underscoring the growing competitive pressure from DeepSeek.

DeepSeek’s rapid ascent over the past three weeks has intensified rivalry not only with global players but also among Chinese tech firms.

“The AI model war is no longer just China versus the US – competition within China is also intensifying as companies like DeepSeek, Alibaba, and others innovate and optimize their models to serve a high-scale domestic market,” said Neil Shah, partner and co-founder at Counterpoint Research. “Chinese companies are being pushed to innovate further due to resource constraints, including limited access to the most advanced semiconductors, global-scale data, tools, infrastructure, and audiences.”

The race for frugal AI

The race to develop high-performance, cost-efficient AI models is intensifying, challenging the business strategies and pricing structures of major US hyperscalers and AI firms as they seek to recover billions in investment.

“This gives enterprise buyers and decision-makers more leverage, increasing pricing pressure on AI applications built with more expensive underlying models,” Shah said. “Such breakthroughs will force enterprises to reconsider, or at least rethink, the economics of AI investments and their choice of models and vendors.”

DeepSeek is driving immediate pricing considerations in two key areas of AI – raw token costs and model development expenses. These factors may force AI companies worldwide to consider optimizing their models to remain competitive. 

“DeepSeek’s success also highlights the power of open source, strengthening the argument that open-source AI could become a dominant market later,” said Hyoun Park, CEO and chief analyst at Amalgam Insights. “If that happens, companies with strong open-source business models for enterprises – such as IBM Red Hat and Canonical – could step in and rapidly scale AI-related managed services.”

The geopolitics advantage

Geopolitics remains a wild card for Western AI firms, potentially tilting the market in their favor by restricting the adoption of Chinese models in certain regions.

At the same time, China is likely to tighten controls on the use of Western AI models, mirroring restrictions seen with other tech applications.

Enterprises will first assess whether these models comply with global privacy and regulatory standards before adopting them at scale, said Sharath Srinivasamurthy, associate vice president of Research at IDC.

“DeepSeek’s advancements could lead to more accessible and affordable AI solutions, but they also require careful consideration of strategic, competitive, quality, and security factors,” Srinivasamurthy said.

However, China’s substantial investment in AI research and development is only beginning to yield results, according to Srinivasamurthy. Other Chinese firms, like Alibaba, which have also been investing in AI in recent years, may soon start launching their own models.

Microsoft first raises doubts about DeepSeek and then adds it to its cloud

January 30, 2025 /

Despite initiating a probe into Chinese AI startup DeepSeek, Microsoft has added the startup’s latest reasoning model R1, to its model catalog on Azure AI Foundry and GitHub.

As part of the blog post in which Microsoft declared that it has added R1, it said that the model had “undergone rigorous red teaming and safety evaluations, including automated assessments of model behavior and extensive security reviews to mitigate potential risks.”

US Government sued after mass emails to federal workforce allegedly sent from insecure server

January 29, 2025 /

When officials working for the incoming Trump administration decided they wanted to email the entire federal workforce last week, they didn’t hang about.

Far from it: A new private class action lawsuit brought by two anonymous US executive branch employees alleges that they simply turned up at the HQ of the US Office of Personnel Management (OPM), which handles HR, and demanded to plug in their email server and get going as soon as possible.

The one person who could have refused authorization for such a move — Melvin Brown II, who took control of the agency’s IT systems only a week before — had already been sidelined.

The suit was filed after OPM sent two test emails to an estimated 2.3 million federal employees in a way that, the suit alleges, broke the E-Government Act of 2002 and was inherently insecure. Those rules require that a Privacy Impact Assessment (PIA) be carried out first.

The day after the suit was filed, the OPM sent another email to federal employees, inviting them to resign.

In addition to its allegations of using an insecure email server, the suit claimed that the person who received the data from the email campaign was a non-OPM employee connected to Elon Musk, raising questions about how any personally identifiable information (PII) arising from it will be stored and secured and whether normal security and procurement protocols were flouted.

Phishing test

On the other side of this campaign were employees who rarely receive mass emails from the OPM’s HR department in a system that normally channels communications through individual agencies.

That might explain why some employees were confused by the unexpected contacts. The first email, which arrived on January 24 from an OPM hr@opm.gov email address, stated that it was testing “a new distribution and response list” designed to allow direct OPM communication with employees. Employees were asked to reply “yes” to the message and asked to visit an OPM website announcing the test.

On January 26 a second email from the same address arrived in inboxes, again asking employees to reply “yes” even if they had already replied to the first email test. With no sense of irony, the message warned employees to be wary of unknown emails:

“As a reminder, always check the ‘From’ address to confirm that an email is from a legitimate government account and be careful about clicking on links, even when the email originates from the government.”

Some employees took them at their word, posting suspicions on Reddit that the emails might be part of a phishing attack or test. It was also noticed that the emails weren’t digitally signed, a standard way of authenticating a sending email server.

“This is EXACTLY how to design a phishing email. Is this a joke? Is this an active cybersecurity operation by a bad actor???,” read one comment.

Walked right in

The employee lawsuit alleges that last week’s emails were part of a wider and hastily assembled campaign to collect data on government employees. 

As part of that, it references a message posted to Reddit by a someone claiming to be an OPM employee with knowledge of the matter, saying that lists compiled from email replies were to be sent to Amanda Scales, an employee who works for Elon Musk and not the OPM.

“Someone literally walked into our building and plugged in an email server to our network to make it appear that emails were coming from OPM. It’s been the one sending those various ‘test’ messages you’ve all seen. We think they’re building a massive email list of all federal employees to generate mass RIF notices down the road,” said a Reddit post referring to reductions in force (layoffs), according to the lawsuit.

Not coincidentally perhaps, this week the OPM emailed a controversial “deferred resignation offer” to all federal employees offering eight months of pay and benefits for anyone who agrees within seven days to resign their positions.

“Type the word ‘Resign’ into the “Subject” line of the email. Hit ‘Send’,” it read. The notice was entitled “Fork in the Road”, perhaps a reference to an artwork of the same name Musk commissioned in 2022.

OPM breach

The OPM, of course, has form when it comes to data security. In 2015, it detected a huge data breach affecting 22.1 million employee records, including PII such as social security numbers. That led to Congressional hearings and several government reports that identified a depressing list of underlying causes.

But with this history in mind, the idea that an unknown party could simply plug their email server into the OPM network without security vetting of either the server itself or its data collection and storage routines will astonish anyone in cybersecurity.

The incident suggests a culture where speed and shock matters above all. It’s not clear how many employees were forewarned that the emails might turn up but asking employees to reply to an email or click on a link is lax in an era of phishing attacks. That’s before considering the possibility that the email server or its data might itself be targeted.

The OPM did not immediately respond to questions sent to the hr@opm.gov email address.

200 UK companies adopt a permanent four-day work week

January 29, 2025 /

Two years after participating in a groundbreaking four-day work week experiment, 200 companies in the UK have chosen to adopt the policy permanently, marking a significant shift in workplace norms.

By continuing the policy, as many as 5,000 employees at those companies will work one day less a week with no reduction in pay.

The pilot program was conducted by the nonprofit 4 Day Week Global, the UK’s 4 Day Week Campaign and Autonomy, a think tank. It guided the companies and their workers through a six-month test of a 32-hour, four-day week. Including a previous study of companies in US and Ireland.

Boston College Professor Juliet Schor, the lead researcher in the 4-Day Work Week study, said she was not surprised the companies continued their truncated schedules, as almost all of the firms in the study reported positive results.

“They are continuing because it is successful,” Schor said. “There are a few reasons for this. Employee well-being goes up a lot. Self-reported productivity goes up even more. So, the companies are getting happier, healthier employees who are typically as productive (or more) than on a five-day schedule.”

The companies that participated in the UK experiment and chose to continue include charities, marketing and technology firms.

Many US and Canadian companies have also adopted four-day work weeks, Schor said. Companies worldwide, including in Ireland, Australia, NZ, Germany, Portugal, and Brazil, have seen similar positive results from trials.

A number of new countries are planning trials and research, as well. Italian, Nordic, French and Belgian trials are already in process, according to Schor. “And I think there are a few more that people are trying to organize,” she said.

In the US, the concept of a four-day workweek is also beginning to germinate. A 2024 Harris Poll for the American Psychological Association found that 81% of 2,027 employed adults believe they could be as effective working four days a week — and 67% think a four-day work week will become the norm in the US during their lifetime.

Additionally, the study found the percentage of US employers offering four-day work weeks rose from 14% in 2022 to 22% last year.

Joe Ryle, campaign director of the 4 Day Week Foundation, argues that the traditional 9-to-5, five-day work week, introduced by Henry Ford in 1926 and the office norm by the mid-20th century, is outdated. Initially, the five-day, 40-hour schedule was seen as a balance between productivity and leisure. However, over time, many have criticized it, calling for more flexibility and better work-life balance.

“We are long overdue an update,” Ryle said. “As hundreds of British companies and one local council have already shown, a four-day week with no loss of pay can be a win-win for both workers and employers. With 50% more free time, a four-day week gives people the freedom to live happier, more fulfilling lives.”

A survey by UK-based Spark Market Research found that 78% of 18- to 34-year-olds expect a four-day workweek to become common in the next five years, while 65% oppose a return to traditional full-time office work.

“This group also say[s] that mental health and improving their overall wellbeing are their top priorities, so a four-day week is a really meaningful benefit and a key enabler of their overall quality of life,” said Lynsey Carolan, managing director of Spark Market Research.

iPhones on T-Mobile can access Starlink satellite communications

January 29, 2025 /

Starlink support appears to be a secret iOS 18.3 feature for iPhones, some of which now support satellite communication services from SpaceX — but only for users invited to join public beta testing of the service on US carrier T-Mobile. This supplements Apple’s existing satellite services supplier, Globalstar.

The stars look different

These are the details as reported by Bloomberg:

  • The service is only available to US users. 
  • T-Mobile’s list of eligible devices for beta testing Starlink/T-Mobiles direct-to-cell services now includes iPhones.
  • Some high-end Samsung Galaxy and Android devices are also being tested.
  • The service is only available to selected iPhone owners running the latest iOS 18.3 update.
  • It will be made available over time to iPhone users on T-Mobile who apply to test the beta service.
  • If enabled, iPhone users can use this instead of Apple’s existing Globalstar service. It is enabled using a new toggle in cellular data to enable satellite messaging.
  • While use of Starlink requires you to manually make a connection, the Starlink integration is always on if it is on at all.

Apple’s existing satellite-based services include Emergency SOS by Satellite, Roadside Assistance via Satellite, Send Location via Satellite and, with iOS 18, Messages via Satellite. T-Mobile Starlink currently supports sending and receiving text messages, location sharing and texting 911 emergency services while connected to Starlink’s satellites, with plans to pursue voice and data coverage next. 

(This support will extend to music and audio podcasts, apparently.)

It is not yet clear whether the new support for Starlink on T-Mobile extends to Apple’s existing services, or if it is confined to those from Starlink. 

Commencing countdown?

The partnership between T-Mobile and Starlink was revealed just months before Apple’s initial introduction of its Emergency SOS via Satellite service with Globalstar.

Apple continues to make serious investments in Globalstar, but the new arrangement with T-Mobile hints that the company might also intend support for Starlink. If it chooses to make that support platform wide, rather than carrier-specific as it is now, then it’s possible iPhone users in more than 100 countries might be able to use the service.

The news comes scant days after Starlink announced its own partnership with Scandinavian Airlines (SAS), which lets passengers access high-speed in-flight internet, even over remote areas such as the North Pole. Oracle is also using Starlink to support some enterprise communications.

Despite the new deal, Apple and Globalstar are not sitting still staring at the Starlink. Just weeks before T-Mobile’s news, an Apple patent appeared in which researchers described a tech that will smoothly shift connections to new satellites as they orbit. That tech could make it possible for the service Apple offers with Globalstar to make and maintain satellite connections more reliably, potentially without pointing your device at the sky. It effectively turns the satellites into the orbital equivalent of mobile network masts. 

Apple invested $1.1 billion in GlobalStar in November 2024, taking a 20% stake in the company in exchange for 85% of the satellite carrier’s network capacity. All this activity, and the détente with Starlink, suggest the race toward satellite-based iPhone communications is intensifying. It’s only a matter of time until global network services become a reality.

Which way to go

If you are an existing T-Mobile customer with a recent iPhone (15/16 series) running iOS 18.3 or later you can join the beta to try Starlink. To do so, you must register here, share your personal details and confirm your registration.  Once registered, “When you are in an area without traditional or roaming cellular services, your satellite-optimized device will automatically connect to the T-Mobile Starlink network if you are part of the T-Mobile Starlink beta,” according to the carrier.

When connected, your device will display “T-Mobile SpaceX” at the top left of the screen. While you can use 911 to contact emergency services, you are unable to send images of videos via the service. You cannot select satellite when other cellular connection options are available.

You can follow me on social media! Join me on BlueSky,  LinkedIn, Mastodon, and MeWe. 

US officials probe China’s DeepSeek AI amid security and privacy scrutiny

January 29, 2025 /

The Chinese AI app DeepSeek has come under intense scrutiny from both the US and European regulators, raising alarms over national security risks, data privacy concerns, and potential intellectual property theft.

The White House confirmed on Tuesday that the National Security Council (NSC) is reviewing the AI model’s implications as fears mount that Chinese advancements in AI could threaten the dominance of US-based AI firms including OpenAI and Google.

“This is a wake-up call to the American AI industry,” White House Press Secretary Karoline Leavitt said responding to reporters during her first press briefing, reinforcing the administration’s commitment to ensuring US leadership in AI.

Leavitt also confirmed during the briefing that she had personally discussed the matter with the NSC.

Meanwhile, Italy’s data protection authority, the Garante, launched its own investigation into DeepSeek, demanding clarity on its data collection practices. The Italian regulator has given DeepSeek and its affiliated companies 20 days to respond, making it one of the first regulatory bodies to take direct action against the Chinese AI startup.

“The Authority, considering the potential high risk for the data of millions of people in Italy, has asked the two companies and their affiliates to confirm which personal data are collected, from which sources, for which purposes, what is the legal basis of the processing, and whether they are stored on servers located in China,” the regulator said in a statement.

The Garante is seeking details about the personal data collected, its sources, its legal basis for processing, and whether any data is stored in China — raising broader concerns over data sovereignty and compliance with Europe’s stringent privacy laws.

Italy’s move comes amid broader concerns about foreign AI models’ compliance with regional regulations. The country had previously banned OpenAI’s ChatGPT in 2023 over potential violations of EU privacy rules, demonstrating its proactive stance in regulating AI models that handle personal data.

“According to its privacy policy, DeepSeek explicitly says it can collect “your text or audio input, prompt, uploaded files, feedback, chat history, or other content” and use it for training purposes,” research firm Forrester said in a statement. “It also states it can share this information with law enforcement agencies [and] public authorities at its discretion.”

Forrester suggested enterprises to “Educate and inform your employees on the ramifications of using this technology and inputting personal and company information into it. Align with product leaders on whether developers should be experimenting with it and whether the product should support its implementation without stricter privacy requirements.”

IP theft concerns deepen in the US

Beyond privacy issues, US officials are also raising alarms over the possibility of intellectual property (IP) theft tied to DeepSeek.

Trump’s AI and crypto policy lead, David Sacks, suggested that China may have leveraged a technique called “distillation,” where AI models learn from other advanced systems—raising the possibility that US-developed AI technology may have been replicated without authorization.

“I think one of the things you’re going to see over the next few months is our leading AI companies taking steps to try and prevent distillation,” Sacks said in a Fox News interview. “That would definitely slow down some of these copycat models.”

“DeepSeek’s ‘open source’ nature opens it up for exploration – by both adversaries and enthusiasts,” said Chester Wisniewski, director and global field CTO at Sophos. “Like Llama, it can be played with and largely have the guardrails removed. This could lead to abuse by cybercriminals, although it’s important to note that running DeepSeek still requires far more resources than the average cybercriminal has.”

During his administration, former President Joe Biden imposed sweeping export controls on AI-related technologies to slow China’s AI progress. Now, Trump’s administration is signaling it may take even stronger measures.

Market reactions and competitive pressure

The emergence of DeepSeek has already rattled the tech industry. On Monday, global investors dumped shares of major US AI companies, fearing the rise of a low-cost Chinese competitor. DeepSeek, which presents itself as a budget-friendly alternative to AI models like OpenAI’s ChatGPT, has quickly gained traction — briefly overtaking ChatGPT as the top AI assistant on Apple’s App Store in the US.

“More pressing for companies, however, is that, due to its cost-effectiveness, we are likely to see various products and companies adopt DeepSeek, which potentially carries significant privacy risks,” Wisniewski added. “As with any other AI model, it will be critical for companies to make a thorough risk assessment, which extends to any products and suppliers that may incorporate DeepSeek or any future LLM. They also need to be certain they have the right expertise to make an informed decision.”

Despite the concerns, Trump framed the development as an opportunity for American firms to step-up their innovation efforts. “The release of DeepSeek AI from a Chinese company should be a wake-up call for our industries,” Trump told House Republicans at their annual policy retreat. “We need to be laser-focused on competing to win.”

Trump also suggested that competition from China could drive American firms to develop AI models at lower costs. “We always have the ideas. We’re always first,” he said, signaling confidence in US ingenuity.

Industry watchers too believe the rise of DeepSeek will accelerate competition.

“DeepSeek’s approach is expected to accelerate the shift toward open-source AI, compelling tech giants to either adapt or risk being left behind,” said  Muskaan Jain, senior analyst at Everest Group. “This shift will likely trigger price wars, faster AI development cycles, and heightened geopolitical tensions over AI dominance.”

What’s next?

With regulatory scrutiny intensifying on both sides of the Atlantic, DeepSeek’s fate in key global markets remains uncertain. If the NSC’s review results in policy recommendations, American AI firms may soon face even stricter export controls and regulations designed to prevent unauthorized AI knowledge transfers.

Meanwhile, DeepSeek’s compliance with European privacy regulations could dictate whether it maintains a foothold in markets like Italy and beyond. If found in violation, the Chinese AI firm could face penalties or access restrictions, echoing past actions against OpenAI.

For US enterprises, these developments signal both risks and opportunities. While competition from China grows, American firms will be under increasing pressure to maintain their technological edge while navigating evolving regulatory frameworks. In an industry moving at an unprecedented pace, policymakers and business leaders alike will need to balance innovation with security and compliance.

“If DeepSeek secures a foothold in the AI market, western firms must respond with responsible innovation and strategic investments to maintain technological leadership while ensuring AI remains ethical and secure in an increasingly fragmented global market,” Jain added.

How Satya Nadella is weathering the storm of Trump’s first days

January 29, 2025 /

It’s going to be a long four years for Microsoft as the company tries to navigate the stormy chaos and pledges of retribution that follow wherever President Donald J. Trump’s mind wanders. Though he’s been president for only a little more than a week, Trump has already unleashed a tsunami of actions that could affect the company: declaring war on diversity (DEI) efforts, removing the guardrails former President Biden placed around AI, and demanding undying fealty as far as his eyes can see.

CEO Satya Nadella has clearly been planning how the company can make it through the years as unscathed as possible. Let’s look at how he’s handled Trump so far — and what he might do over the next four years.

Avoiding AI wars

Microsoft is the world’s largest and most valuable AI company, and one of Trump’s primary goals for AI is to unleash its power, for better or worse. So, his actions will have outsized effects on Microsoft.

Trump wasted no time getting to the “worse” part — on his first day in office, he rescinded former President Biden’s executive order putting guardrails around AI, including safety and security measures, as well as issues related to bias, civil rights, and how generative AI (genAI) is produced. 

Microsoft and many other AI companies welcomed Biden’s order when it was issued in 2023, because it was good for tech companies, not just for people affected by AI. With guardrails in place, Microsoft and other companies believed more people would use AI, which would mean bigger profits for them. 

After killing Biden’s order, Trump signed his own, asking that an AI action plan be developed by the feds in the next 180 days. The order is as much an attack on Biden as it is anything else — its first bullet point called Biden’s move a “Dangerous Executive Order that hinders AI innovation and imposes onerous and unnecessary government control over the development of AI.”

We don’t know what the Trump Administration plan will be. But it’s clear it won’t target the dangers of AI. Microsoft, as well as other AI companies, have remained silent about the action. It was a smart move for Nadella to hold his tongue — Microsoft can always impose its own guardrails on AI, despite Trump’s order. 

Nadella has also stayed on the sidelines in the controversy around Trump’s Stargate project, a joint initiative by OpenAI, Oracle, Softbank, and the US government to raise $100 billion to build AI data centers and power plants, and then up the ante all the way to $500 billion. Top Trump tech advisor Elon Musk claimed OpenAI, Oracle, Softbank “don’t actually have the money” for the project, called OpenAI CEO Sam Altman a “swindler,” and said the entire initiative was “fake.” 

Altman fired back, telling Musk he was “wrong, as you surely know” and taking another shot at Musk, writing, “just one more mean tweet and then maybe you’ll love yourself.”

Nadella sidestepped the fight when asked about it in an interview on CNBC, saying that Microsoft was spending $80 billion on Azure so “customers can count on Microsoft with OpenAI models being there everywhere in the world, serving OpenAI models and other models. I am good for my $80 billion,” he added.

Again, this is a smart move. No matter what happens with Stargate, Microsoft wins. It’s a big OpenAI investor, so if Stargate succeeds, it pays off for Microsoft. And if it doesn’t, Microsoft can sell more AI-related infrastructure. Joining the tit-for-tat wouldn’t have helped Microsoft.

Staying the course on DEI

Trump was elected in part because of his attacks on DEI efforts — and since he’s become president, he’s gone on an anti-DEI rampage. Not only has he shut down the federal government’s DEI programs, he’s also threatened to investigate companies that continue to pursue DEI goals.

Many tech titans joined Trump’s crusade, notably Meta’s Mark Zuckerberg, who killed Meta’s DEI efforts, and Jeff Bezos, who scaled back Amazon’s. Google has downsized its DEI program as well.

Microsoft hasn’t followed suit. Nadella in October said in Microsoft’s annual report on DEI that the efforts “ensure our work force represents the planet we serve, and that the products we build always meet our customers’ needs.” 

In December, more than a month after Trump’s election, Microsoft Chief Diversity Officer Lindsay-Rae McIntyr doubled down on that in a post on LinkedIn, noting that not only are DEI efforts the right thing to do, but they’re essential to Microsoft’s success. She wrote on LinkedIn, “I’m thinking about the importance of continuing our diversity and inclusion work, expanding empathy, and anticipating the needs of all our stakeholders, both within Microsoft and beyond. The business case for D&I is not only a constant, but is stronger than ever, reinforcing our belief that a diverse and inclusive workforce is crucial for innovation and success.”

Once again, Nadella handled this in a low-key way, promising the company will stay true to its values, but not going out of his way to take on a fight with Trump about it.

Not singing in Trump’s amen chorus

Beyond individual issues like AI and DEI is the larger issue of pledging fealty to Trump and all he stands for. Trump has made tech one of the cornerstones of his plans for the future and is using his relationships with tech moguls as a way to provide cover for the ways in which he wants to reshape the country and its politics.

Many of the world’s most powerful tech titans are only too happy to sing along in Trump’s amen chorus. To show their support, Zuckerberg, Bezos, Google CEO Sundar Pichai, and Musk were all seated in prominent positions at Trump’s inauguration. (Note that Apple CEO Tim Cook attended as well, but Apple still fully supports DEI efforts.)

Sen. Elizabeth Warren (D-MA) wrote: “Big Tech billionaires have a front row seat at Trump’s inauguration. They have even better seats than Trump’s own cabinet picks. That says it all.” 

Noticeably absent was Nadella. His non-attendance made clear he’s not in on Trump’s agenda.

The upshot

What we’ve seen in the earliest days of the Trump administration from Nadella is likely how he’ll try to safely navigate Microsoft through the coming years. He’ll stay out of the public eye as much as possible, won’t comment on political and cultural issues, and when it comes to tech won’t engage in the kind of chest-banging that Musk, Zuckerberg, Altman, and other tech titans have engaged in.

All that is easy to say and tough to do. But if Nadella can do it, Microsoft will be a stronger company four years from now, with its culture intact.

3 exceptional text expansion tools for Windows

January 29, 2025 /

Do you find yourself typing the same things over and over?

If the answer is yes — and if you type enough text during the workweek, it almost certainly will be — I’d highly recommend you install a text expansion application. These text shortcut tools are a must-have productivity upgrade for many knowledge workers, and they can even help you at home. If you haven’t given one a go yet, you’re missing out.

There are am wide variety of text expanders to choose from. Many are paid tools intended for businesses, complete with pricey subscriptions and features that only make sense for large teams and complex professional workflows.

(These types of apps are particularly relevant in the PC arena; text shortcut tricks are built into Android phones and Apple devices, but on Windows, you need a third-party application to take advantage of them.)

I’ll cover the text expanders I recommend — easy-to-use, free solutions for Windows PCs. If your workplace pays for a different one, that’s great. Otherwise, I suggest starting with one of the options here.

Want more Windows PC tips in just a few minutes each week? Sign up for my free Windows Intelligence newsletter. You’ll get free Windows Field Guide downloads as a special welcome bonus!

Beeftext: A free Windows text expander

Up first is Beeftext, a completely free, open-source, and as-simple-as-it-gets text expansion tool for Windows PCs. My colleague Jared Newman recommended it at PCWorld — and for good reason.

After installing and launching Beeftext, you’ll see a little red “Bt” icon in your system tray. Open it to get started.

In Beeftext, combos are text expansion shortcuts. You’ll click Combos > New to create a new one. Then, you’ll just need to enter a Snippet and a Keyword. When you type the keyword in any application, Beeftext will replace it with the snippet.

For example, you might want to try creating the keyword /add and specifying your address. Then, you can type that keyword in any application on your PC to quickly insert your address.

BeefText
While creating a combo, you just need to fill in the “Snippet” and “Keyword” boxes.

Chris Hoffman, IDG

It’s really that simple. You could enter /# and associate that with your phone number. You don’t have to use a slash — it’s just an easy way to create memorable shortcuts that you won’t type accidentally.

You can do a lot with this type of tool. If you often find yourself responding to a customer, client, or even just a friend with a specific message, you might create a snippet for that message so you can quickly type it in emails. If you often end up inserting some kind of signature or disclaimer message, you might create a snippet for that. It’s really up to you.

Autocorrect: Built right into Office apps

It’s worth noting that Office apps like Microsoft Word, Excel, OneNote, and Outlook have a text-expansion-style feature built right into them. It will only work inside those Office apps. But, if you only need this feature while working in those specific applications, it’s perfect. And it might be the only solution for workers who can’t install anything else on their PCs.

First, open the application you want to use — such as Microsoft Word or OneNote. Click the “File” menu and select “Options.”  Click the “Proofing” option in the left sidebar, and then click the “AutoCorrect Options” button.

The AutoCorrect window will appear. Under “Replace text as you type,” add something you want to replace at the left and the text you want to replace it with at the right. Then, click the “Add” button to add it to the list, and click “OK.”

For example, as in the above example, you could enter /add and have Office replace that text with your address when you type it.

AutoCorrect in Word
AutoCorrect is incredibly convenient. I wish it worked elsewhere in Windows, too!

Chris Hoffman, IDG

Want to use formatted text — or text with multiple lines? Type some text in your document, and then select it with your mouse. Then, click File > Options > Proofing > AutoCorrect Options. The text you selected will be prefilled there, and you can associate a text shortcut with it.

Text Blaze: An extension for Chrome and Edge

Last but not least, Text Blaze is another exceptional option for Windows-based text replacement. (We use this application at our newsletter-focused small business The Intelligence.) While it’s not completely free, it does offer a free plan for individuals. That plan gives you up to 20 snippets, which is plenty for many people.

Text Blaze installs right in modern browsers like Chrome and Edge as a browser extension. (Unfortunately, it doesn’t work with Firefox.) It’s a speedy and simple way to quickly replace text in your browser — I heartily recommend it.

Text Blaze dashboard
Text Blaze lets you specify a “shortcut” that will be replaced by a “snippet” when you type It on any web page in your browser.

Chris Hoffman, IDG

On the business front, the tool has some useful features that work well for our team, such as sharing snippets with other people and dynamically entering information in snippets. That’s more than many users will need, though. The paid subscription won’t be necessary if you just want some quick text shortcuts in Google Chrome for your own use.

Text Blaze does have a Windows application, too — but, in my experience, it works better and faster as a Chrome extension.

More text expansion options for Windows PCs

With all the text expansion tools available, it’s hard to narrow the list. If you do some web searches, you’ll see a lot of other similar options. Here are a few other popular possibilities and what you need to know about them — including why they didn’t earn one of my top recommendations:

  • Espanso is a completely free and open-source text expander that works on Windows as well as other platforms. However, it doesn’t have a convenient graphical interface, so you’ll have to write out your “matches” and “replaces” in a text editor. If that works for you, go right ahead — it will be the tool of choice for many geeks.
  • TextExpander is incredibly polished and feature-filled. But this tool doesn’t have a free version. While it might be a great fit for some businesses who want to use the sharing functionality with their team, the $40/year subscription here isn’t a great fit for the average individual.
  • Phrase Express is another application with a long history. But, starting at $100 per license, it’s probably not the ideal choice for most people.

I can’t emphasize enough: There are many other text expansion tools that’ll work with Windows. Lots of them likely work well, too, but most aren’t free or have other issues that make them not-entirely-optimal choices for most common purposes.

Bonus: Pin text to Clipboard History in Windows

Don’t want to install a text expander? You can also “pin” items to your clipboard history in either Windows 11 or Windows 10. To open it, press Windows+V and choose to enable clipboard history if you haven’t already done so.

Then, after copying some text or an image to your clipboard, you can press Windows+V and use the “pin” option to save it to your clipboard history for later. You can then press Windows+V in the future and select a pinned item to quickly paste it.

Clipboard history with pinned items
The Clipboard History tool is useful for both text and images.

Chris Hoffman, IDG

It’s not a text expander, exactly — but it is a way to quickly insert some saved text. And it’s that end result that counts the most.

There’s more where this came from: Sign up for my free Windows Intelligence newsletter! You’ll get three new things to try every Friday and free in-depth Windows Field Guides.

The AI bust is here

January 29, 2025 /

Hands up! Who remembers the dot.com bust? 

I do. It was March 10, 2000. The NASDAQ Composite index peaked at 5,048.62 points. That doesn’t sound like much today, but it was up 800% from 1995. Then people started to get nervous about the true value of the dot.com businesses driving the market, interest rates started to rise, and by October 2002, the NASDAQ had fallen 78% from its peak. 

We’re not even a month into 2025, and the new Chinese AI program DeepSeek this week sparked Nvidia’s $465 billion rout, the biggest single-day drop in US stock market history. 

This is just the beginning. It’s not that DeepSeek is so much better than ChatGPT’s OpenAI or any of the other popular generative AI (genAI) tools. Maybe it is, maybe it isn’t. (Just don’t ask it about what happened at Tiananmen Square in 1989.) 

No, what matters is that DeepSeek requires an order-of-magnitude less computing power to achieve similar results. By programmer Simon Willison’s count, “DeepSeek v3 trained on 2,788,000 H800 GPU hours at an estimated cost of $5,576,000. For comparison, Meta AI’s Llama 3.1 405B (fewer than DeepSeek v3’s 685B parameters) trained on 11x that; it required more than 30.8 million GPU hours and was trained on 15 trillion tokens of data.

What does all that mean? As Larry Dignan, editor-in-chief of Constellation Insights, explained, “You can get API access to DeepSeek’s R1 model for 14 cents for a million tokens compared to OpenAI’s $7.50.” 

In other words, “large language models (LLM) pricing is going to collapse.

“The scary part is the LLM giants didn’t have profit margins to begin with,” he continued. “LLMs are going to commodity in a hurry.” That’s bad news for companies such as OpenAI. Sure, the company’s market value stands at $157 billion — but it’s still losing billions of dollars. If OpenAI’s customers decide they don’t want to pay its rates when they can get much the same service for a fraction of the cost from DeepSeek, where does that leave OpenAI?

What happens then? Look at it this way: today’s overheated stock market is driven by the Magnificent Seven— companies that with one exception (Tesla) have one thing in  common: they’re all heavily invested in AI hype. Even Tesla is heavily invested in what Elon Musk calls real-world AI. 

It’s not just the Magnificent Seven; pick-and-shovel AI companies such as Oracle, Super Micro, and Nebius (formerly Yandex) are also in a world of trouble.

If anyone can build LLMs without a ton of money, why exactly is Nvidia worth trillions? Couldn’t someone disrupt Microsoft, Meta, or Google? Indeed, Perplexity’s AI chatbot is already better than Google at search. It’s not just people like me for whom search is part of what they do for a living; businesses are also turning to Perplexity from Google.

Bratin Saha, DigitalOcean’s chief product and technology officer, sees the potential for a relatively small company to become a big player. “DeepSeek AI is the ‘Android moment for AI’ (given that ChatGPT was likened to the ‘iPhone moment for AI’), as it shows groundbreaking AI products from the open-source community,” Saha told me.

“DeepSeek AI democratizes AI and cloud computing because it shows you do not need multi-billion dollar investments for compelling innovation,” he said. “It lowers the barrier for small and medium enterprises and individual developers to work with AI.” 

Exactly. 

The way Stephen O’Grady, co-founder and industry analyst with RedMonk, sees it, enterprises have two major AI concerns. How trustworthy the technology is and its cost. As O’Grady points out, “Enterprises have been shocked, in many cases, at the unexpected costs – and unclear returns – from some scale investments in AI.”

He’s got that right. O’Grady continued, noting that DeepSeek challenges some core assumptions by enterprises. These are: 

  • What if enterprises don’t have to rely on closed, private models for leading-edge capabilities?
  • What if training costs could be reduced by an order of magnitude or more?
  • What if they don’t require expensive, state-of-the-art hardware to run their models?

This changes everything for businesses that want to use AI and multi-billion dollar companies that want to sell AI to you. It’s a sea change.

“What the markets reacted to was DeepSeek’s ability to build a model that rivaled OpenAI’s sophisticated o1 reasoning model and Anthropic’s Sonnet model for pennies on the dollar on a fraction of the compute,” Jim Zemlin, executive director of the Linux Foundation, wrote. “It also appears that DeepSeek did this using well-known techniques. There were no massive algorithmic breakthroughs, just very clever engineering.”

DeepSeek, he noted, uses open-source software. Yes, technically, part of it isn’t open source, but one of DeepSeek’s models is under the MIT License. As Stefano Maffulli, the Open Source Initiative (OSI) executive director, pointed out, DeekSeek’s models also fail to be open.

Still,”a small team in China took a fresh look at a problem and came up with a novel approach that reduced the cost of chain-of-thought reasoning by 50x (if DeepSeek’s postings are accurate) and then published a paper fully describing their process, allowing the community to benefit from their learnings,” Zemlin said. “We need MORE of this progress, not less.… This is a struggle over open markets between the forces of open and the forces of closed.”  

That may be good for open source and for AI, but for our stock market, which is largely driven by AI, it’s another matter. The Magnificent Seven accounted for over 50% of the S&P 500’s gains in 2024. If they collapse, a good chunk of the market will follow.

I’m not normally a bear, but I’ve seen this rodeo before. Oh, AI will survive, and eventually, it will prove useful and profitable. But, in the meantime, I can see an AI crash coming, and it won’t be pretty.