Ke-Quang Nguyen-Phuc

💥 Linkup annonces a 3M€ funding round to plug LLMs to high-quality content.    Much like the music industry before Spotify, AI applications today face hurdles in accessing quality content. In the early 2000s, people often turned to platforms like Pirate Bay to download music, but that came with legal risks, poor sound quality, and artists missing out on fair compensation. Similarly, today many AI tools depend on existing web frameworks like search engines and web scraping to collect information.   However, as the AI landscape evolves, these methods are proving less efficient and raising new concerns around legal compliance and fair use of copyrighted content.   Linkup aims to solve this problem by taking a new approach with premium content providers with both a legal and ethical consideration.   Delighted to support this top-notch team Philippe Mizrahi, Denis Charrier, Boris Toledano alongside Seedcamp, Motier Ventures, Kima Ventures OPRTRS CLUB and amazing BAs ! cc Sia Houchangnia, Nicolas Essayan, Christophe Raynaud, Noé Gersanois, Chloe Timsit, Alix François-Poncet, Axeleo Capital (AXC)

118

5 commentaires

Outbrain acquires Teads merge for a whopping $1 billion! Perfect example of how consolidation is reshaping the adtech landscape ! 🚀 🌟 Why is this a smart move for Outbrain? 🎥 They now have a solid video solution 💼 Teads brings a strong advertiser-focused approach 💰 Higher wallet share from premium publishers 🔗 Brand+ Performance $ It's crystal clear: For big adtech companies to stay competitive and scalable, they must hit that $2-3 billion revenue sweet spot and embrace multi-format solutions. 📈💡💰. 📊🔍💼 This is especially good news for the supply side because publishers need stronger and fewer alternatives for monetization rather than working with 40+ partners. Huge congrats to my friends at Teads! 🎉 who have worked hard to get here. What do you think this means for the future of adtech? Share your thoughts below! 👇 https://lnkd.in/gtkvHH2H #AdTech #Acquisition #DigitalAdvertising

5

Is #Nvidia Still a Buy📈 and Should You Include It In Your Portfolio 💼 ??? Nvidia reported strong earnings📊 last week but the stock went down 📉! So what happened??? Don't you worry, we put everything in context for you to make informed decisions. Join us in this week's Market Buzz 🎙️where we delve into Nvidia earnings, possible risks and our outlook looking forward. Subscribe to the channel to get your weekly dose of news, market opinions from Crypto to fundamental stock picking 👇 https://lnkd.in/g5jMbKgq #eToro #stocks #technology #risk #money #podcast #collaboration #opinion #forecast #politics #geopolitics #semiconductor #microsoft #apple #amazon #magnificent7 Timothy Assi Mimi Ho Akim Guerreiro (Etcaetera) Adamos Hadjiantoniou, ACA, CFA, ACSI Nestor Mendoza

12

1 commentaire

MSR has showcased BitNet, a novel 1-bit Transformer architecture, designed to address the challenges associated with scaling LLMs. A key innovation in BitNet is the introduction of BitLinear, which replaces traditional linear layers and facilitates the training of 1-bit weights from the ground up. This architecture not only achieves competitive performance in language modeling but also significantly reduces memory usage and energy consumption when compared to advanced 8-bit quantization methods and FP16 Transformer baselines.   The architecture employs quantization-aware training, optimizing the model for low-precision representation from the outset, which enhances both accuracy and stability during the training process. BitNet features stacked blocks of self-attention and feed-forward networks with binarized weights, while retaining higher precision for other components, thereby minimizing computational costs. This design allows for efficient matrix multiplication with 1-bit weights, resulting in notable energy savings, especially as model sizes increase.   The training methodology incorporates techniques such as the straight-through estimator for gradient approximation and mixed precision training. These approaches enable BitNet to converge effectively with larger learning rates compared to traditional FP16 Transformers, leading to improved training efficiency and stability, which is particularly beneficial in resource-constrained environments.   Comparative evaluations indicate that BitNet consistently outperforms existing quantization methods across a range of benchmarks, including zero-shot and few-shot tasks, while also maintaining lower inference costs. These findings highlight the advantages of quantization-aware training over post-training quantization, positioning BitNet as a prominent approach for developing efficient LLMs.    Arxiv: https://lnkd.in/e-D959-U

3

Good products start with sketches Removing detail raise the signal/noise ratio for product discussions. That's how we started Slite back at eFounders studio in 2016 with Pierre, Thibaud, Quentin. With simple mock ups, no details, purely speaking of the key components. How will the editor interact, what were the fundamental organisation features, how comments would fit in, could you decorate your space etc. I kept this approach over the years, drafting and discussing each key product as sketches first. It also became a great way to delegate to a bigger team. A good wireframe is enough to convey the essential, while letting freedom of execution and ownership. And today the Hexa team released the internal tool they use to do these, "Blocks" on Product Hunt 🐈 🎉 You can see below the example of our latest (unreleased) product AskX built with Blocks, it's extremely simple and a pleasure to use! Go give it a try and give them some love on PH! (first link in the comment)

38

4 commentaires

Interesting insights for startups planning to expand business to US.

5

Crunch Lab, a startup from our portfolio, has just raised $3.5M in a seed round led by Multicoin Capital ! Big big congrats to Jean & Arnaud and the whole team for the amazing work. I witnessed impressive dedication from the whole team to close this round and reach new heights. What does Crunch Lab la do? Crunch Lab makes it easier for big companies to unlock the value in their data by running machine learning competitions through their community, CrunchDAO, which has over 6,000 data scientists, quants, and ML engineers. These experts compete to build AI models that give companies actionable insights, mainly to the finance industry. With this funding, Crunch Lab will ramp up their operations and dive into new data-rich markets. Bravo! Link in the comments for the full article.

34

5 commentaires

"Agentic AI" is the buzzword in VC. But can AI agents actually plan and act ? 💡 The idea is that AI systems can plan and execute workflows autonomously with goals inputs. Example: "book me a 7 day trip in France that costs no more than X to discover hidden cultural gems". Does it work with the current AI systems? 🙅‍♂️ According to DeepMind co-founder Demis Hassabis, current AI systems are not good at planning, they can't problem solve, they can't do actions in the world for you. They are "very much passive Q&A systems". However according to Demis, this might be the next era in AI. 👀 By combining the work done with AlphaGo in games (agent players that are very good at planning) with multi-modal "real world" learning (Gemini) and applying this to language and everyday workloads, he believes that we might get there. João Duarte Darko Butina Xavier Mesnier Pepijn Rot

21

14 commentaires

There a fewer and fewer public companies.  They were +8000 in the US in 1996 and are now only 4642.  That’s down 43%. How did this happened? You can explain this in many ways :  The regulation makes it increasingly difficult for a company to IPO VC & PE have be on the rise for the last decade Acquisitions are also on the rise. To give you a snapshot : 870 publics companies were acquired by one of the magnificient seven (google, apple, Meta, Amazon, Microsoft, Tesla and Nvidia) And what’s the consequence? More and more companies, facing difficulties by IPO are drawn to VC & PE were there are more and more investors. Even though the private market lack transparency and available data, VC and PE firms have been working on tools to obtain data in ordre to make better investment décision. It’s the case for Earlybird or EQT Ventures. And also a lot of start-ups that are now working to create tools to help VC in their decision making process. So now the question is : will private markets one day have a sufficient data level for investors to make as informed decision as they would in public markets ? On my side I truly believe that in the upcoming years, we'll see the first quants VC. If you want to read some more : https://lnkd.in/eKFZuyeE

6

I am thrilled to announce our new strategic collaboration with NVIDIA 🤝. In the era of Siliconomy, leveraging next-generation GPUs is a game changer 🌟. It’s crucial for building advanced AI/GenAI solutions, from in-store analysis to digital twinning, consumer engagement, and interactive avatars or AI agents. By combining strategic insights and world-class AI consulting services, and as a proud partner of NVIDIA, we promise our clients the maximum potential of NVIDIA’s industry-leading GPUs and solutions, such as NeMo and Riva. We are already working on multiple use cases, including risk management, smart stores, and e-learning AI agents. More than ever, Sia Partners is ready for the hypertransformation of our world. Let’s deliver it together! 🚀 Serge Palaric Matthieu Gasse-Hellio Ségolène Ramadier Elsa Martineau David Martineau #ArtificialIntelligence #MachineLearning #Innovation #DataScience #SiaPartners #NVIDIA #DigitalTransformation.

575

9 commentaires

A must read for every founder (and investor) in B2B marketplaces. ⤵️ All credits to Fabrice Grinda + FJ Labs 👏🏽 https://lnkd.in/eB7uGzyH #firstdegree #founders #investors #whyrelationshipsmatter
#alwaysbelearning #sharingknowledge

1

When I founded Hexa (eFounders) with Thibaud Elziere back in 2011, the term Startup Studio hadn't been coined yet; it emerged around 2014. In 2019, #TeamEntrepreneurship was termed to showcase how we operate as a studio amidst the growing number of Studios. At Hexa, we've established 14 core principles, our "Hexa Cheat Sheet," reflecting our values and mission for the next decade. Fueled with Ambition and Driven by Purpose Discover more in the full article & pdf in the first comment.

134

7 commentaires

Tomorrow I’m lucky to be hosting a panel with three exceptional founders: - Gilles Gaillard - Co-founder of Animaj, raised 100m EUR (in equity and debt) in 2023 to build an AI-powered animation studio that counts 𝗮 stunning 15 billion (!) YouTube views per year. Animaj acquires animation and cartoon franchises to turn them into digital-first content and amplifies their distribution using AI. - Baptiste Jamin - Co-founder of Crisp IM, 42 alumni, also can be called the “King of Bootstrap”. He started building chatbots in 2015, 𝗯𝗲𝗳𝗼𝗿𝗲 𝗶𝘁 𝘄𝗮𝘀 𝗰𝗼𝗼𝗹. Crisp centralises all customer interactions through live chat, chatbots, and a shared inbox. Today Crisp serves over 600,000 companies without raising any funds. - Vincent SIMON - Co-founder of Relationchips and founder of Simundia. Vincent is a serial entrepreneur & investor who founded a coaching platform that raised over 15m EUR & is currently working on a no-code sales workflow platform. What do they have in common? They started their career as freelancers and worked their way up to become highly successful entrepreneurs. 👉 Join us at Le Wagon tomorrow at 7pm to hear their story. We’ll also host a networking session for you to meet people from our community who are working on projects that might interest you. 🔗Link to register in the comments.

41

2 commentaires

Time to step up Germany 🇩🇪 🇫🇷 Corporations have pledged to invest €15bn in France at the “Choose France” summit — an annual event held since 2018 to facilitate foreign investments in the country. A large chunk of the money will be dedicated to AI. Microsoft stole the show after announcing its largest investment in France to date, a €4bn cash envelope to build more data centres — and expand existing ones — across the country by the end of 2025. The focus is on increasing the capacity of GPUs, which are needed to run large AI models. Amazon also announced that it’ll invest €1.2bn to further develop its French cloud and AI infrastructure. Quantum computing was also on the agenda. IBM pledged to invest €45m to expand the company’s research efforts in France. Up to 50 quantum researchers and engineers will be recruited in the country in the next year.

39

4 commentaires