This holiday season, millions of shoppers are set to buy gifts online, but tech companies are vying to make AI agents the new shopping assistants. Platforms like Perplexity, OpenAI, and Google are developing AI tools that can browse websites, select products, and even complete purchases. Perplexity recently launched a shopping agent that combines navigation and checkout features, though it’s still ironing out inefficiencies.
AI-driven shopping isn’t without challenges. Early tests show agents struggling with stock availability and delayed purchases, while companies like Perplexity rely on human oversight to address errors. Privacy concerns are also emerging, especially with AI systems accessing billing information. However, partnerships like Perplexity’s with Stripe, which uses single-use payment cards, aim to mitigate risks and provide secure transactions.
These tools could revolutionise online shopping by saving time and uncovering hidden deals, but they also threaten traditional e-commerce models. Retailers and advertisers may resist as fewer consumers visit storefronts and targeted ad opportunities shrink. Despite the hurdles, 2025 is expected to see significant advancements in AI shopping agents, promising a glimpse into the future of effortless online retail.
AI hardware startup Tenstorrent has secured a $693M Series D funding round, valuing the company at over $2.6B. The investment, led by Samsung Securities and AFW Partners, includes participation from Hyundai and Bezos Expeditions, among others. Founded in 2016 and based in Toronto, Canada, Tenstorrent aims to challenge Nvidia’s dominance in the AI chip market.
Tenstorrent’s CEO, Jim Keller, a renowned microprocessor engineer, announced plans to develop AI training servers and expand its engineering team using the new capital. The company has also committed to releasing a new AI processor every two years, with signed customer contracts amounting to nearly $150M. This move positions Tenstorrent among a growing number of startups racing to innovate in AI hardware, alongside competitors such as Axelera, Etched, and Groq.
The funding highlights escalating investor interest in alternative AI chipmakers as demand for cutting-edge computing solutions soars. With its ambitious roadmap and backing from high-profile investors, Tenstorrent is poised to carve out a significant share of the burgeoning AI hardware market.
Cancer drug development faces a pressing challenge: most new compounds fail to advance through clinical trials, despite rising cancer rates, particularly among younger adults. French entrepreneur Fanny Jaulin believes the root cause lies in outdated trial designs, not the drugs themselves. Her startup, Orakl Oncology, founded in 2023 as a spinoff from the Gustave Roussy Institute of Oncology, aims to revolutionise this process by blending data analysis with biological insights.
Unlike competitors focusing solely on AI or biology, Orakl combines the two to tackle cancer’s complexity. The approach leverages organoids—miniature, simplified organ versions—to test drug responses, supplemented by patient avatars that integrate tissue samples with extensive datasets. These datasets, though smaller than those of some competitors, contain 40 variables per patient, allowing a focus on hard-to-treat cancers like colorectal and pancreatic.
Orakl plans to commercialise two products: O-Predict, which forecasts how patients might respond to drug candidates, and O-Validate, designed to match drugs to biological data. These innovations cater to pharmaceutical developers and biotech firms alike, supported by nearly €15 million in funding, including a recent seed round led by Singular.
Jaulin’s mission goes beyond business success. She seeks to address the therapeutic gaps in precision medicine, making drug discovery faster and more effective. With cancer increasingly becoming a chronic condition, Orakl’s ultimate goal is to bring more life-saving treatments to patients.
Nebius Group has secured $700 million through a private placement, attracting investors such as Nvidia, Accel, and Orbis Investments. The AI infrastructure firm, founded by former Yandex CEO Arkady Volozh, aims to enhance its capabilities to serve artificial intelligence developers globally.
The funding will enable Nebius to accelerate its investments in GPU clusters, cloud platforms, and other AI development tools. Having already committed $1 billion in investments by mid-2025, the firm hinted at potential further expansion. With more than half its clientele based in the United States, Nebius is leasing data centre space in Kansas City, Missouri, and exploring additional growth opportunities.
As part of the placement, Nebius issued 33,333,334 Class A shares at $21 per share, reflecting a slight premium to recent Nasdaq trading averages. The financing was oversubscribed, leading to a revised annualised revenue projection of $750 million to $1 billion by the end of 2025.
Nebius also announced it would no longer pursue a previously approved share buyback, citing strong investor interest and favourable market conditions. Chairman John Boynton stated that shareholders who wished to exit had ample opportunity to do so at competitive prices.
Amazon is developing an AI-designed material aimed at reducing carbon emissions in its data centres. Created by Orbital Materials, the carbon-filtering substance acts like a sponge on an atomic level, selectively capturing CO2. The material represents a step towards Amazon’s goal of achieving net-zero carbon emissions by 2040.
Orbital Materials claims the new material could add just 10% to hourly GPU training costs for AI, significantly cheaper than traditional carbon offsets. Data centres increasingly demand more energy and water for cooling, posing sustainability challenges. Amazon Web Services (AWS), the world’s largest cloud-computing provider, plans to trial the innovation in 2025 at a single facility.
The partnership with Orbital spans three years, aiming to explore additional AI-designed materials for water and cooling needs. Orbital uses AI to simulate and develop these substances, with the material being synthesised in its lab established last year. The collaboration also includes making Orbital’s open-source AI tools available to AWS customers.
Orbital, co-founded by Jonathan Godwin, operates from Princeton and London and is backed by Radical Ventures and Nvidia’s venture arm. Godwin, formerly with DeepMind, emphasised the importance of sustainable innovations in addressing growing environmental concerns linked to AI-driven energy demands.
Heathrow Airport, one of the world’s busiest, is trialling an advanced AI system named ‘Amy’ to assist air traffic controllers in managing its crowded airspace. Handling nearly half a million flights annually, Heathrow aims to improve safety and efficiency through real-time data and advanced tracking capabilities provided by the AI system.
Amy integrates radar and 4K video data to give controllers a detailed visualisation of aircraft positions, even when out of sight. Designed by NATS, the UK’s air traffic management agency, the system offers vital information such as flight numbers and aircraft types, helping controllers make faster, more informed decisions. After testing on over 40,000 flights, NATS plans to fully operationalise a ‘digital contingency tower’ by 2027 to ensure backup in emergencies.
Despite its promise, experts caution against over-reliance on AI. They highlight potential limitations, such as insufficient contextual judgment and challenges in handling unexpected scenarios. Colin Rigby from Keele University emphasised that AI should complement human operators rather than replace them.
The adoption of similar AI-driven solutions is being explored by major airports worldwide, including those in Singapore, New York, and Hong Kong, signaling a shift toward digital transformation in air traffic management.
A team from Johns Hopkins and Stanford has trained robotic systems to perform surgical tasks with human-like precision. Using a da Vinci Surgical System, the researchers applied ‘imitation learning,’ where robots observe recorded surgical videos to replicate complex movements like suturing and tissue manipulation. This innovative method eliminates the need for manual programming and allows robots to learn from the combined expertise of skilled surgeons.
The AI-powered system combines imitation learning with advanced machine learning techniques, enabling it to convert visual data into precise robotic actions. Not only does it perform surgical tasks proficiently, but it can also self-correct in real time, such as retrieving a dropped needle without human intervention. Such adaptability could reduce complications and enhance patient outcomes.
This breakthrough accelerates the path toward autonomous robotic surgery. Researchers believe robots can now learn new procedures in days rather than months. While full autonomy in surgery remains a future goal, this advancement marks a significant step toward safer and more accessible healthcare worldwide.
Ann Jensen, a woman from Salisbury, was deceived into losing £20,000 through an AI-powered investment scam that falsely claimed endorsement by UK Prime Minister Sir Keir Starmer. The scammers used deepfake technology to mimic Starmer, promoting a fraudulent cryptocurrency investment opportunity. After persuading her to invest an initial sum, they convinced her to take out a bank loan, only to vanish with the funds.
The scam left Ms. Jensen not only financially devastated but also emotionally shaken, describing the experience as a “physical reaction” where her “body felt like liquid.” Now facing a £23,000 repayment over 27 years, she reflects on the incident as a life-altering crime. “It’s tainted me for life,” she said, emphasising that while she doesn’t feel stupid, she considers herself a victim.
Cybersecurity expert Dr. Jan Collie highlighted how AI tools are weaponised by criminals to clone well-known figures’ voices and mannerisms, making scams appear authentic. She advises vigilance, suggesting people look for telltale signs like mismatched movements or pixelation in videos to avoid falling prey to these sophisticated frauds.
The PongBot, a cutting-edge AI-powered tennis robot, is revolutionising solo tennis practice. Unlike traditional ball machines, this smart device uses a clip-on sensor to track your court position and adjusts its shots in real-time. Whether you’re a beginner or a seasoned player, it simulates match conditions by varying ball speed, spin, and placement, offering a dynamic training experience anytime.
Players can personalise their sessions via the PongBot app, which features up to 300 preprogrammed drills and custom sequences. With speeds of up to 80 mph and an eight-hour battery life, the AI robot provides intense training tailored to skill level. Advanced AI match training makes it feel like playing against a real opponent, pushing players to improve their game.
Initially funded through a wildly successful Kickstarter campaign, the PongBot comes in two models priced at $699 and $899. With seamless integration with smart devices like Apple Watch and support from tech giant Qualcomm, this high-tech training partner is set to ship in December, perfect for off-season practice.
French startup Linkup is reshaping how AI applications access web content by creating a marketplace for licensed material. Unlike traditional web scraping, Linkup partners with publishers to fetch content directly through integrated systems, ensuring intellectual property rights are respected. The platform caters to developers enriching large language models (LLMs) with high-quality, fresh data.
CEO Philippe Mizrahi highlighted Linkup’s focus on licensing deals that benefit both publishers and AI developers. The service targets business applications, such as corporate insights or sales tools, utilising databases like Statista and news sources.
With €3 million in seed funding, Linkup aims to expand its team and services, standing out in the growing market of ethical content acquisition for generative AI. Competitors like ScalePost also focus on licensing, indicating a shift in how AI firms source data amid tightening regulations.
