- AI Leadership Weekly
- Posts
- AI Leadership Weekly
AI Leadership Weekly
Issue #20
Welcome to the latest AI Leadership Weekly, a curated digest of AI news and developments for business leaders.
Top Stories

Source: YouTube
Framework releases AI-empowering desktop
Hardware company Framework recently had their "2nd Gen" event where they unveiled several product refreshes and announced two entirely new products. The most interesting new product is their Framework Desktop, powered by AMD Ryzen AI Max+ chips.
On their own, they are comparable to a NUC or Mac Mini/Studio. And, like all Framework products, they can be bought as a complete unit (with a case, M.2 SSD, etc.), or just as a mainboard. But, what has us most excited for the AI space is their ability and focus on running large AI models.
In the live keynote (linked above), they demonstrate linking together multiple motherboard units which they say can power the full DeepSeek R1 model. While these feats aren't new, Framework is doing it at a significantly lower cost (more than half the cost of a Mac Studio), all while being on open and standard hardware.

Klarna moves back to human customer service
A little over a year ago, Swedish fintech company Klara reported on their efforts to shift their chat support away from human workers and onto AI chatbots. In their first months, they boasted at AIs handling 2/3rds of their chat volume.
Today, though, they are adjusting their "strategy" to bring back more human-human interactions. In a statement, Klarna said they're moving towards a more "balanced approach" where AIs handle routine inquiries, while humans will handle the harder stuff. They also highlight humans handling "high-value customer interactions."
In a tweet, Klarna CEO Sebastian Siemiatkowski said that cost savings from using chatbots will help fund "even better" human service.

Source: Anthropic
Hybrid reasoning with Claude 3.7 Sonnet
Anthropic has announced what they are calling the first "hybrid reasoning" model.
They say that Claude 3.7 Sonnet (which we'll just call "Sonnet") can give either quick, near-instant responses like we’re used to with older models or typical ChatGPT requests, and that it’s also capable of the slow and step-by-step results of the new reasoning models. API users can also specify how long they want the model to 'think' for.
In Brief
Market Trends
Chegg suing Google over AI search summaries
In a first, the online education company Chegg has sued Google over their AI search summaries.
The now-ubiquitous results which appear at the top of Google (and many other) results pages go further than the previous 'summaries', which quoted directly from sites and took from structured data, and uses an LLM to create a custom (and, often, incorrect) summary of search results.
Chegg claims that this feature is hurting traffic to their site and, in turn, their revenue. This is a sentiment already expressed by SEO and content marketers who rely on website content ranking highly in Google to attract visits and conversions. A Google spokesperson, however, says that the features sends "billions of clicks" every day to a wider ranges of websites.
Orders for Nvidia H20 chips surge in China
Outlets are reporting a "boom" of Chinese companies ordering Nvidia's H20 AI chips, and comes in response to companies there jumping on the DeepSeek bandwagon.
There had been fears of a chip-demand downturn in recent weeks when DeepSeek R1 was first announced, and was claimed to have been trained on significantly less hardware than what is being used in the US. This recent news of further H20 orders may indicate otherwise.
Apple joining the $500 billion bandwagon
In what sounds like familiar news, Apple has announced intentions to invest $500 billion over the next four years to build out AI data centres and infrastructure within the US.
This will include a giant facility in Texas, where they can assemble AI servers, as well as more investments into TSMC's Arizona chip plant. Their master plan also includes adding 20,000 research jobs across the country, and expanding their domestic manufacturing efforts.
Tools and Resources
Wan2.1
Try out the new video- and image-generator from Alibaba (which they claim beats OpenAI's Sora!). Fair warning: it’s currently experiencing the hug of death.
EasyGen
Write your LinkedIn posts using this AI tool, and get to networking!
SmolVLM2
A small model which can analyse videos on your laptop and other lop-powered devices.
Recommended Reading
What’s after LLMs?
Computerphile discusses where AI research will go once LLM tech plateaus.
How to run DeepSeek locally
We're all more and more concerned about privacy and foreign (and domestic?) governments getting our data. This short guide will show you how to get started experimenting with DeepSeek on your local machines, be that at home or in the office, and far from prying eyes!
Hit reply to let us know which of these stories you found the most important or surprising! And, if you’ve stumbled across an interesting link/tweet/news story of your own, send it our way at [email protected] It might just end up in the next issue!
Thanks for reading. Stay tuned for the next AI Leadership Weekly!

Brought to you by Data Wave your AI and Data Team as a Subscription.
Work with seasoned technology leaders who have taken Startups to IPO and led large transformation programmes.