Generative AI for Business # 39

Here are your weekly key insights and tools on Generative AI for business, covering the latest news, strategies, and innovations in the B2B sector and their business impact.

While we move forward with our lives (and this newsletter), the devastating LA wildfires have left many families and communities with nothing. This newsletter is a place to connect about work, but I know many of us are asking, “How can I help?”

I’m stepping outside the boundaries of work to share ways you can provide direct assistance to those whose lives and livelihoods have been disrupted by the fires. Below are resources if you want to help—and please feel free to share or repost to spread awareness.

Los Angeles Fire Department Foundation: donations https://supportlafd.kindful.com/

World Central Kitchen: serving free meals to evacuees and firefighters across LA: https://wck.org/relief/california-fires-jan25?

American Red Cross: shelter, food, and medical supplies for all: https://lnkd.in/gma-Hi9Z

GoFundMe: verified fundraisers of families in need: https://www.gofundme.com/c/act/wildfire-relief/california#/?

Animal Wellness Foundation: taking animals in need of shelter: https://www.animalwellnessfoundation.org/

Pasadena Humane: helping animals in need of shelter: https://give.pasadenahumane.org/give/654134/#!/donation/checkout

Thank you,

Eugina

Models

Despite massive investments, LLMs like Databricks' DBRX, Snowflake's Arctic, and NVIDIA’s Nemotron-4 flopped in 2024, while Meta’s Llama thrived by being free, strategic, and developer-friendly. Perplexity introduced ads post-response in its LLM results but risks low CTRs unless it integrates ads more seamlessly into user interactions. Cohere’s North platform simplifies enterprise AI integration with secure, air-gapped deployments, already gaining traction with the Royal Bank of Canada. Elon Musk’s Grok broke free from X (Twitter) with a standalone app, offering real-time web access and privacy-first features to challenge ChatGPT. Meanwhile, DeepMind’s Genie showcases the power of pre-built AI systems, generating physics-aware 3D worlds and proving that leveraging existing models is the smarter path to innovation.

• LLM is a costly business unless you know how to get people to use it. LLMs that Failed Miserably in 2024? Despite the race to dominate large language models (LLMs) heating up, some contenders crashed and burned in 2024. Databricks' $10M DBRX saw only 23 downloads on Hugging Face, despite claims of surpassing GPT-3.5. Snowflake’s Arctic, Stability AI’s Stable LM 2, and AMD's OLMo also struggled with minimal adoption. Meanwhile, Falcon 2 and NVIDIA’s Nemotron-4 failed to keep up with fast-evolving rivals like Meta’s Llama 3.3. Even AI21 Labs’ Jamba, boasting a 256K token context window, barely captured attention. In a year of lofty promises, these models highlight the brutal reality of a competitive LLM landscape.

My take: Some LLMs thrive while others fizzle because success isn’t just about big budgets or flashy features, it’s about strategy, ecosystem, and execution. Meta’s Llama models, for instance, succeed because they’re free (well, mostly), instantly winning developer loyalty. On top of that, Meta integrates its models into a thriving ecosystem of products, creating seamless use cases. Compare that to Databricks’ DBRX or Snowflake Arctic, which, despite millions in investments, lack both the buzz and a clear narrative for their utility. Meta also understands the power of marketing; Llama isn’t just a model; it’s a trend. By contrast, Databricks and Snowflake seem to have missed the memo that even great tech doesn’t sell itself. Lastly, Llama models are purpose-built and focused, while some competitors come off as solutions in search of a problem. In the end, it’s not just about building a mode, it’s about making it matter.

• Perplexity Tries Out Ads In Its LLM Results Perplexity has introduced ads within its language model (LLM) results, marking a step toward monetizing its free services. Ads appear in the "Related" section after search results and are labeled as "sponsored." These ads are absent in the paid version. CEO Aravind Srinivas had recently discussed the potential for AI agents to manage ads directly, allowing companies to bid for recommendations made by the AI, streamlining user interactions. While Perplexity's current ad placement strategy differs from Google's, with ads appearing after answers instead of before, it signals progress in monetization efforts. As one of the first LLMs to adopt advertising, Perplexity could influence how AI advertising evolves in the future.?

My take:? Perplexity is in a unique position to pioneer this new ad model, creating a potential competitive advantage. However, the current implementation may lack the visibility and immediacy of traditional ad placements like Google’s search ads as this is what we are used to. While Perplexity's approach avoids interrupting the user experience, placing ads post-response risks lower click-through rates (CTRs) and reduced advertiser ROI. The "Related" section might not draw enough user attention, limiting the model’s ability to scale revenue. Additionally, advertisers will demand measurable ROI before committing budgets to AI agent-based advertising, especially when compared to established platforms like Google and Meta.

To succeed, Perplexity must enhance ad relevance by seamlessly integrating ads into the user journey, offering actionable insights and solutions that add immediate value, such as dynamically appearing within answers as "suggested tools" or "recommended resources" to boost utility and visibility. It also needs to develop transparent metrics to provide advertisers with robust data on ad performance, such as conversions driven by AI-assisted interactions, ensuring clear ROI. As a pioneer in introducing ads within LLMs, Perplexity must educate the market by defining and demonstrating the value of this paradigm, showing how AI agent-driven bidding systems can outperform traditional methods. Finally, given the rapidly evolving AI advertising space, Perplexity must iterate quickly, experimenting with ad formats, placements, and targeting strategies to refine its go-to-market approach. If executed correctly, Perplexity has the chance to become the "Google Ads" of the AI era. However, the journey from experimentation to market leadership will require strategic foresight, iterative development, and bold innovation in both AI capabilities and advertising models. The next 12 months will determine whether Perplexity can move beyond being an early mover to setting the gold standard for AI monetization. Or Meta might be winning again like it’s winning in the LLM wars.

• North: The AI Platform Where Work Gets Done | Cohere Cohere has launched North, a secure AI workplace platform designed to simplify the integration of its Command large language models into business workflows. North offers a chatbot interface for tasks like analyzing earnings reports, finding documents, and automating HR processes. Users can create AI agents tailored to specific tasks without needing programming expertise. Powered by Cohere’s Compass search tool, North uses advanced AI models to extract and rank relevant data from documents, slides, and spreadsheets. It supports deployment in both cloud and on-premises environments, including air-gapped systems for highly regulated industries. Early adopters include the Royal Bank of Canada, with potential for industry-specific versions in the future.

• xAI's standalone Grok iOS app launches in the US — here's how to find it | Tom's Guide xAI, Elon Musk's AI startup, has launched a standalone iOS app for its AI chatbot platform, Grok, which was previously tied to the X app (formerly Twitter). The app is now available in the U.S., Australia, and India. Grok offers a conversational tone, real-time web access, image generation, and privacy-focused outputs. Unlike before, users no longer need an X account to access Grok, broadening its appeal. With features comparable to ChatGPT and Gemini, Grok’s release positions it as a strong competitor in the AI chatbot space. An Android version is expected, though no release date has been announced.

• Google's DeepMind is recruiting AI researchers to advance world model development - SiliconANGLE? DeepMind’s focus on advancing world models like Genie highlights the strategic value of leveraging pre-built AI systems and fine-tuning them with specialized data. Genie’s ability to generate interactive, physics-aware 3D worlds from text shows immense potential for applications in robotics, media, and real-time planning. By building on foundational models rather than creating from scratch, DeepMind maximizes scalability and efficiency—a strategy I advocate for industries like telecom: utilize existing AI models, train them on proprietary data, iterate, and focus resources on solving real-world problems. This approach accelerates innovation, reduces costs, and drives tangible ROI.

News?

NVIDIA dominated CES with Project Digits, a $3,000 AI supercomputer designed to democratize AI development, alongside RTX AI PCs running localized generative AI and the Cosmos platform enabling Physical AI training in digital environments. Samsung unveiled Vision AI, turning TVs into adaptive smart companions with Neo QLED 8K visuals, generative wallpaper, and SmartThings integration, aiming to enrich daily life. Qualcomm expanded its Snapdragon X chips to mid-range Windows laptops and desktops, delivering multi-day battery life and AI acceleration at affordable prices. Meanwhile, AMD shook up gaming and AI workloads with its Strix Halo Ryzen AI Max+ processors, boasting revolutionary memory tech and outperforming rivals like Intel Lunar Lake and NVIDIA RTX 4090. Together, these announcements signal an industry racing to merge AI, hardware, and user experiences for the next wave of innovation.

• NVIDIA made announcements last week in conjunction with CES 2025. Here are the key ones:

What is actually really cool, that this supports a new “law” introduced for GPUs: Huang's law - Wikipedia? Huang’s Law, named after Nvidia CEO Jensen Huang, observes that GPU performance doubles every two years, significantly outpacing Moore’s Law for CPUs. First articulated at Nvidia’s 2018 GPU Technology Conference, Huang highlighted GPUs’ superior advancement—achieving 25x faster performance in five years compared to CPUs’ tenfold increase under Moore’s Law.

This acceleration is attributed to the synergy between hardware, software, and AI, as well as the GPUs’ parallel processing architecture. For example, Nvidia GPUs reduced AI training times from six days (2010) to 18 minutes (modern DGX-2 systems). Critics argue the law relies on Moore’s Law’s foundation and question its long-term viability. However, Huang’s Law underscores the transformative potential of GPUs in advancing AI and computing efficiency.

• NVIDIA Launches AI Foundation Models for RTX AI PCs At CES 2025, NVIDIA unveiled foundation models for RTX AI PCs, powered by GeForce RTX 50 Series GPUs with FP4 compute capabilities. These GPUs enable generative AI to run locally with enhanced performance and reduced memory needs. NVIDIA NIM microservices provide preconfigured AI workflows for digital humans, content creation, and productivity tasks, while AI Blueprints offer user-friendly reference workflows for developers. The Llama Nemotron open models support complex AI tasks like coding and chat. NIM-ready RTX AI PCs will be available from major manufacturers starting February 2025, marking a significant step in accessible, localized AI development.

?

• Cosmos World Foundation Model Platform for Physical AI | Research NVIDIA has launched the Cosmos World Foundation Model Platform, a tool designed to aid developers in creating customized world models for Physical AI applications. The platform includes a video curation pipeline, pre-trained world models, tools for fine-tuning models, and video tokenizers. These capabilities allow Physical AI systems to simulate and train in digital environments before real-world deployment, making them more adaptable to various applications like robotics and autonomous vehicles. NVIDIA has made the platform open-source, with open-weight models available under permissive licenses via NVIDIA Cosmos, aiming to address critical global challenges through accessible innovation.

?

• ?Now See This: NVIDIA Launches Blueprint for AI Agents That Can Analyze Video NVIDIA has unveiled a new AI Blueprint for video search and summarization, powered by its Metropolis platform and advanced technologies like Cosmos Nemotron vision models and NeMo Retriever. This tool enables organizations to build AI agents capable of analyzing vast quantities of video data at 30x real-time speed, addressing issues such as productivity loss, safety risks, and operational inefficiencies in industries ranging from manufacturing to sports. Applications include quality control, safety monitoring, and asset management in industrial settings, as well as enhancing player analytics and fan engagement in sports. Global partners like Accenture and Deloitte are already integrating the blueprint into their workflows, with early access now available.?

• Nvidia CEO: AI Advancing Self-Driving Cars, Robotics, Digital Manufacturing at CES 2025 (summary) At CES 2025, NVIDIA unveiled foundation models for RTX AI PCs, powered by GeForce RTX 50 Series GPUs with FP4 compute capabilities. These GPUs enable generative AI to run locally with enhanced performance and reduced memory needs. NVIDIA NIM microservices provide preconfigured AI workflows for digital humans, content creation, and productivity tasks, while AI Blueprints offer user-friendly reference workflows for developers. The Llama Nemotron open models support complex AI tasks like coding and chat. NIM-ready RTX AI PCs will be available from major manufacturers starting February 2025, marking a significant step in accessible, localized AI development.?

My take. If you head is spinning of all the announcements, you are nto alone, let me break it down and then tie it all together for you. It’s a cohesive strategy that integrates hardware, software, and AI to redefine how we interact with technology across industries. From Project Digits, a $3,000 AI supercomputer democratizing AI development for researchers and universities, to foundation models for RTX AI PCs that enable localized generative AI with cutting-edge GPUs, NVIDIA is lowering barriers to AI innovation. The Cosmos Platform advances "Physical AI" by enabling robots and autonomous systems to train in digital twin environments, while the AI Blueprint for video analysis enhances productivity and safety with 30x faster video processing. Tied together by Huang's Law—GPU performance doubling every two years, these announcements highlight NVIDIA’s unified vision of agentic AI, where hardware and software synergize to solve real-world challenges, accelerate innovation, and make AI accessible for everyone, from startups to industry giants.

They’re not throwing darts at random AI applications—they’re engineering a synchronized ecosystem. Project Digits democratizes hardware, foundation models and NIM microservices unlock localized AI, Cosmos prepares physical AI systems for the real world, and Metropolis video agents close the loop by solving real-world problems. It’s not hardware here, software there—it’s all interwoven.

In other words, NVIDIA isn’t building products—they’re architecting the future of AI itself. And if Jensen Huang’s predictions are anything to go by, we’re in for an AI revolution that doesn’t just compute but thinks, reasons, and acts.

• Samsung news from CES shows a similar strategy of software and hardware interplays targeted towards consumer market: Samsung Electronics Unveils Samsung Vision AI and New Innovations at First Look 2025, Delivering Personalized, AI-Powered Screens To Enrich Everyday Life At CES 2025, Samsung introduced Samsung Vision AI, which integrates personalized, AI-powered features across its widest lineup of TVs and displays, including Neo QLED, OLED, QLED, and The Frame. Vision AI transforms screens into adaptive companions, enhancing entertainment and seamlessly integrating with the SmartThings ecosystem for smarter living. Features like Click to Search, Live Translate, and Generative Wallpaper redefine interactivity, while tools like Home Insights and Pet and Family Care ensure security and convenience. Samsung also launched its flagship Neo QLED 8K QN990F, boasting advanced AI-powered visuals and sound, as well as The Frame Pro, which combines art and entertainment with premium Neo QLED picture quality and expanded Samsung Art Store offerings. Innovations extended to the Premiere 5 triple-laser projector with interactive touch capabilities and the MICRO LED Beauty Mirror, blending personalized beauty insights with cutting-edge display technology. Partnerships with Microsoft and Art Basel further highlight Samsung's commitment to reimagining the role of screens in daily life.

Hardware and chips other announcements

• https://techcrunch.com/2025/01/06/qualcomm-brings-its-snapdragon-x-chips-to-more-affordable-windows-laptops-and-desktops/ Qualcomm introduced its Snapdragon X chip series at CES 2025, targeting mid-range Windows laptops and desktops. Built on a 4nm fabrication process, the Snapdragon X offers multi-day battery life, enhanced performance, and AI acceleration through an integrated neural processing unit (NPU). Supporting Bluetooth 5.4, Wi-Fi 7, and up to three external 4K monitors at 60Hz, these chips aim to make AI-powered PCs more accessible at a price point of around $600. The new processors will power Microsoft's Copilot+ PCs and are expected in devices from brands like Acer, Asus, Dell, HP, and Lenovo starting in Q1. Qualcomm reported a 3X increase in native Windows apps for Snapdragon in 2024, signaling growing compatibility despite lingering software challenges. Qualcomm is also expanding into desktops, with Snapdragon X Series-powered mini and tiny PCs set to launch later in 2025, offering improved power efficiency and NPU-accelerated features for developers. This move highlights Qualcomm's intent to challenge AMD and Intel, despite capturing only 0.8% of the PC market in Q3 2024.

• AMD’s beastly ‘Strix Halo’ Ryzen AI Max+ debuts with radical new memory tech to feed RDNA 3.5 graphics and Zen 5 CPU cores | Tom's Hardware? At CES 2025, AMD unveiled its Strix Halo Ryzen AI Max+ processors, designed for high-performance gaming laptops and AI-powered workstations. These chips, powered by Zen 5 CPU cores and RDNA 3.5 graphics, boast advanced shared memory technology that allows up to 128GB system memory, optimizing performance for AI and gaming workloads. The flagship model, Ryzen AI Max+ 395, features 16 cores, 32 threads, and a 40-core GPU, delivering 1.4X faster gaming performance than Intel's top Lunar Lake processor and 2.2X better AI workload performance compared to Nvidia's RTX 4090 at 87% lower power consumption. The new chips support adaptive memory allocation, enabling the GPU to access up to 96GB for AI tasks, significantly reducing memory bottlenecks. With configurable TDPs ranging from 45W to 120W, the processors are tailored for thermally constrained devices. AMD plans to release these processors across Q1 and Q2 2025, with systems from HP, Asus, and others expected to feature these chips. Despite the impressive claims, AMD's real-world gaming benchmarks and availability remain key factors for broader market impact.

Partnerships

• Novo Nordisk, Valo Health Ink Expanded Up-to-$4.6B AI Collaboration Novo Nordisk and Valo Health have expanded their AI collaboration to develop up to 20 treatments for obesity, type 2 diabetes, and cardiovascular diseases. The deal, valued at up to $4.6 billion, leverages Valo's Opal Computational Platform to analyze patient data, identify novel drug targets, and accelerate preclinical drug discovery. The collaboration builds on their 2023 partnership focused on cardiovascular diseases and includes a $190 million upfront payment, R&D funding, and milestone-based payments. The integration of AI-driven insights with wet lab simulations aims to revolutionize human-centric drug development and expand Novo Nordisk's pipeline of cardiometabolic treatments.?

Regulatory?

The tech industry is rallying against the Biden administration’s proposed "Export Control Framework for Artificial Intelligence Diffusion," warning it could stifle U.S. AI leadership by limiting global chip sales, raising costs, and ceding dominance to competitors like China, while calling for a more measured rulemaking process. Meanwhile, the U.S. Department of Health and Human Services (HHS) unveiled its AI Strategic Plan, emphasizing responsible AI use to enhance healthcare, human services, and public health through innovation, equity, and safeguards, while addressing risks and fostering AI-empowered workforces.

• Tech group urges US to halt rule that would limit global access to AI chips | Reuters The Information Technology Industry Council (ITI) has urged the Biden administration to reconsider its proposed "Export Control Framework for Artificial Intelligence Diffusion," which would regulate global access to AI chips. ITI, representing tech giants like Amazon, Microsoft, and Meta, warned that these restrictions could harm U.S. leadership in AI by ceding market opportunities to competitors and increasing costs for domestic companies. Critics, including Oracle, argue the rule is overly broad, potentially hindering the global cloud computing industry. The proposed rule aims to balance national security concerns, particularly regarding China's military advancements, but faces backlash for its rushed implementation and significant economic implications. As the rule's release looms, industry leaders advocate for a more deliberate approach to protect innovation and market competitiveness.

My take: The tech industry is pushing back against the proposed "Export Control Framework for Artificial Intelligence Diffusion" because it risks over-regulating AI chip exports, potentially ceding global market dominance to competitors like China while raising costs and stifling innovation. Companies argue that the rule’s sweeping scope would harm U.S. leadership in AI by imposing constraints on global sales and commercial cloud computing without sufficient stakeholder input. They are advocating for a proposed rulemaking process to allow for deliberation and adjustments. While the Biden administration may want to prioritize national security by restricting AI advancements in adversarial nations, the significant industry backlash, coupled with potential economic and geopolitical ramifications, makes the rule's approval uncertain in its current form, likely requiring revisions before implementation.?

• HHS Releases Strategic Plan for the Use of Artificial Intelligence to Enhance and Protect the Health and Well-Being of Americans The U.S. Department of Health and Human Services (HHS) has released its AI Strategic Plan, which provides a framework for responsibly leveraging artificial intelligence to enhance healthcare, human services, and public health. The Plan outlines four primary goals: catalyzing health AI innovation, promoting trustworthy and ethical AI use, democratizing AI technologies for equitable access, and fostering AI-empowered workforces. With a focus on safety, equity, and access, HHS aims to accelerate scientific breakthroughs, improve care delivery, and optimize public health while addressing risks and maintaining accountability. This dynamic approach includes ongoing updates, risk assessment, stakeholder engagement, and robust safeguards to ensure ethical AI implementation for the benefit of all Americans.?

Regional Updates

Artificial intelligence is set to reshape Asia's labor markets, with advanced economies like Singapore facing higher disruption risks (50% of jobs) compared to emerging economies like Laos (3%), while policymakers must tackle inequalities through reskilling, safety nets, and ethical AI regulations. Meanwhile, Microsoft announced a $3 billion AI investment in India, adding a fourth data center, training 10 million people, and fostering entrepreneurship through partnerships like SaaSBoomi, as it positions India as an "AI-first" nation and competes with tech giants like Amazon for dominance.

• How Artificial Intelligence Will Affect Asia’s Economies Artificial intelligence is expected to significantly affect Asia's labor markets, with advanced economies seeing higher exposure. About 50 percent of jobs in advanced economies are at risk of AI disruption, compared to 25 percent in emerging and developing economies. In Singapore, 40 percent of jobs are highly complementary to AI, while in Laos, this figure is only 3 percent, highlighting regional disparities. Within countries, service, sales, and clerical roles, often held by women, face higher displacement risks, while managerial and technical jobs are more likely to benefit from AI integration. Policymakers can address these inequalities by implementing effective social safety nets, reskilling programs, and education initiatives that help workers adapt to AI-driven changes. Governments can also mitigate risks by enacting regulations that ensure ethical AI use and protect data, promoting inclusive economic growth.

• ?Microsoft plans $3 billion AI investment in India, Nadella says | TechCrunch? Microsoft announced a $3 billion investment to enhance AI and cloud services in India, expanding its footprint in the country with plans to establish a fourth data center region by next year. The company also aims to train an additional 10 million people in AI skills and foster AI adoption among startups and businesses. As part of its efforts, Microsoft signed an agreement with SaaSBoomi to promote entrepreneurship in smaller cities, targeting $1.5 billion in venture funding for Indian AI and SaaS startups. Satya Nadella emphasized Microsoft’s commitment to making India an "AI-first" nation, citing the rapid diffusion of AI across industries. Notable Indian clients like Infosys, Air India, and Meesho are already leveraging Microsoft’s technologies, including Microsoft 365 Copilot, which has significantly improved efficiency for some clients. This move reinforces Microsoft's competition with other tech giants, such as Amazon, which also has significant investments in India.

Investments

We have two subsections in today’s newsletter with a focus on investment in compute power and startup investments. AI investments in 2024 and 2025 are reshaping the landscape, divided into Compute Power Investments and AI Startup Investments. On the compute side, Microsoft leads with an $80 billion AI data center push, including $3 billion in India and $1.5 billion in the UAE, while AWS's $11 billion Georgia expansion emphasizes localized AI and cloud growth. Hyperscale data centers are doubling in capacity, with major players like AWS, Microsoft, and Google projected to dominate 60% of global capacity by 2029, reflecting the surge in GenAI-driven demand. On the startup front, Databricks secured $10 billion, OpenAI $6.6 billion, and Anthropic reached a $60 billion valuation, signaling robust enterprise AI growth, while European players like Wayve and Mistral also saw billion-dollar funding rounds. The divide highlights two crucial pillars of AI's future: infrastructure to power innovation and startups driving application-specific breakthroughs.

Compute power investments

• Microsoft to capture 'golden' AI data center opportunity with $80 billion investment Microsoft has announced an $80 billion investment in AI-enabled data centers by 2025, with over half allocated to the U.S., signaling a strategic push to solidify American leadership in AI. The company aims to enhance AI infrastructure, skill development, and global AI exports, positioning American AI as a competitive alternative to China's offerings. This includes expanding Azure cloud and AI capacity globally, such as a $3 billion investment in India and $1.5 billion in UAE-based G42. Microsoft Vice Chair Brad Smith highlighted the urgency for the U.S. to act swiftly in the global AI race, emphasizing international influence through rapid technology deployment.?

• GenAI is lighting a fire under hyperscale DC capacity growth | TelecomTV? According to Synergy Research Group, the average capacity of new hyperscale datacenters over the next four years will nearly double, fueled by demand for GPU-intensive infrastructure. Major players like AWS, Microsoft, and Google are leading the charge, currently holding 41% of global datacenter capacity—a figure expected to climb to 60% by 2029. Datacenter investments soared in 2024, hitting $282 billion, with public cloud infrastructure dominating the market. GenAI has not only accelerated adoption but also reshaped priorities, with companies retrofitting existing datacenters and expanding globally. As public cloud investments continue to grow, enterprises are also making a comeback, reflecting a vibrant future for datacenter technology.

• AWS says it'll invest 'at least' $11B to expand data center infrastructure in Georgia | TechCrunch Amazon Web Services (AWS) announced plans to invest at least $11 billion in Georgia to expand its data center infrastructure, aiming to support cloud computing and AI technologies. The initiative is expected to generate 550 jobs in the state, building on AWS's previous $11 billion investment in Indiana that created 1,000 jobs. Georgia’s growing appeal for data centers is attributed to its low-cost electricity, robust fiber-optic infrastructure, and state tax incentives. However, local opposition highlights concerns about competing real estate needs and environmental impacts. Driven by AI’s computing demands, data center expansions are set to continue, with AI projected to account for 19% of power demand by 2028.?

AI Startup investments

• The Largest AI Startup Funding Deals Of 2024? The 2024 AI startup funding landscape saw record-breaking investments, with several companies raising billion-dollar rounds that highlight the sector's explosive growth. Databricks secured the largest funding of $10 billion at a $62 billion valuation, enabling further AI product development and expansion. OpenAI followed with $6.6 billion, valuing the ChatGPT creator at $157 billion, marking it as one of the most valuable private companies globally. Elon Musk's xAI raised two $6 billion rounds in 2024, accelerating its generative AI ambitions. Waymo received $5.6 billion from Alphabet, furthering its autonomous vehicle advancements, while Amazon invested an additional $4 billion in Anthropic, making AI a cornerstone of its cloud strategy. Other notable rounds included Anduril Industries and G42 ($1.5 billion each), alongside CoreWeave and Wayve ($1.1 billion each). These investments underscore the ongoing demand for AI innovation across sectors, positioning these startups as key players in shaping the future of AI applications.

• The year of AI: 12 events that shaped the sector in 2024 | Sifted European AI investment hit $11 billion, nearly doubling 2023’s figures, with major players like Wayve, Mistral, and DeepL raising $1 billion+ rounds. Wayve’s $1.05 billion Series C set a European record, while Mistral achieved a €5.8 billion valuation, emerging as a key competitor to U.S. tech giants. Meanwhile, Germany’s Helsing secured €450 million to advance AI-powered defense solutions, and DeepL reached a $2 billion valuation, cementing its lead in AI translation. However, turbulence accompanied the highs—Stability AI and H Company faced leadership challenges, and Aleph Alpha pivoted away from building large language models due to revenue struggles. Regulatory developments like the EU AI Act introduced compliance pressures, while AMD’s $665 million acquisition of Silo AI underscored international interest in European AI innovation.

• Beyond The Hype: AI, Innovation And Rational Investment In 2025 The AI industry continues to mature in 2025, with promising innovation tempered by a more rational investment approach. According to Rob Biederman of Asymmetric Capital Partners, the market is bifurcating: truly impactful AI companies are flourishing, while overhyped startups face reckoning. This divergence reflects broader trends—vertical integration strategies are gaining traction, with firms like Biederman’s buying small businesses to embed tech solutions rather than selling software alone. Simultaneously, legacy startups from the 2020-2021 funding frenzy are running into valuation resets, highlighting the risks of oversaturation and inflated expectations. With venture capital recalibrating toward focused, high-return investments, 2025 could mark a return to strategic, value-driven funding, where craftsmanship in innovation overshadows unchecked growth.??

• What Anthropic’s $60 Billion Valuation Reveals About Enterprise AI’s Next Generation | PYMNTS.com Anthropic’s recent $60 billion valuation underscores the intensifying competition in enterprise AI, where companies like OpenAI, Google, and Anthropic race to offer smarter, secure, and customizable solutions. This rapid evolution highlights the shift from experimenting with generative AI to deploying it at scale across industries like healthcare, legal services, and finance. As businesses prioritize integration capabilities and data security, providers that cater to these needs—like Anthropic's Claude, which focuses on tailored models for enterprise use—are gaining significant traction. The market dynamics reflect a broader trend where AI’s transformative potential drives enterprise spending, which surged sixfold in 2024, reshaping operational efficiency and business growth strategies.

• AI is moving inside the operations, and behind the CEO desk, at America's small businesses Cohere, an enterprise-focused AI company valued at $5.5 billion after its $500 million funding round in July 2024, is targeting industries like healthcare, banking, and IT with its AI agent platform, North. North, which debuted in January 2025, allows users with any level of technical expertise to deploy and customize AI agents for business processes such as HR, finance reporting, and IT support. Unlike competitors like OpenAI and Anthropic, Cohere is doubling down on the enterprise market, bypassing the consumer space. This strategy aligns with the growing demand for secure, customizable AI in regulated industries, where privacy and compliance are paramount. Cohere’s efficient use of capital, achieved by focusing solely on enterprise solutions, has enabled it to maintain operational stability despite rising GPU costs. Supported by investors like Nvidia, AMD, Salesforce, and Oracle, Cohere differentiates itself with its ability to automate specific workflows while fostering long-term recurring revenue. This focus has allowed the company to thrive in a competitive AI landscape where rivals like OpenAI and Anthropic command valuations of $157 billion and $60 billion, respectively.??

Research?

A Fortune 1000 survey shows generative AI adoption is accelerating, with 24% of companies now using it at scale compared to 5% last year, though measuring productivity gains remains elusive. Cognizant research predicts AI will reshape 90% of jobs in the next decade, automating technical tasks while increasing demand for creativity and strategic thinking, urging leaders to balance automation with human interaction. Bosch’s Tech Compass 2025 highlights AI’s global acceptance, with China and India leading adoption, as Bosch trains 65,000 employees in AI applications like HR automation and manufacturing optimization. Intelligent agents, powered by foundation models, are emerging as transformative tools for tasks like data analysis and business automation but face challenges like latency, cost, and security. Meanwhile, an IBM survey reveals significant barriers in generative AI development, including a skills gap, inadequate tools, and a lack of standardized frameworks, despite the growing use of AI coding assistants to improve productivity.

• Study probes trends around AI in the enterprise | CIO? A new survey of Fortune 1000 AI and data leaders reveals that generative AI adoption is accelerating, with 24% of companies now using it at scale, up from just 5% last year. Early-stage production has also grown, with 47% of organizations now testing the technology, compared to 25% in 2024. Most leaders (58%) attribute AI’s value to productivity gains, though few companies measure these improvements carefully. Despite optimism, challenges remain. The role of Chief Data Officer (CDO) is evolving but characterized by high turnover and short tenures. Generative AI has yet to transform organizational data cultures significantly, and leaders remain divided on whether AI initiatives should fall under business or technology leadership.

• Gen AI will transform 90% of jobs in the next decade – how should business leaders prepare? - HR News? According to Cognizant research, generative AI will impact 90% of jobs over the next decade, reshaping roles rather than replacing them. Data-heavy jobs like administrative support and data entry are most vulnerable to automation, while roles such as financial analysts and senior executives are leveraging GenAI for tasks like trend analysis and scenario planning. The demand for creativity, strategic thinking, and communication skills is expected to grow as AI takes over technical tasks. Entry barriers for some professions may also lower, enabling less experienced workers to perform at higher levels using AI-driven tools. Business leaders are urged to embrace GenAI strategically, balancing automation with human interaction. The success of this transition depends on upskilling employees, integrating AI thoughtfully, and preserving human-centric strategies where they add value.

• Bosch Tech Compass Report Released at CES 2025; Focus on AI Skills Bosch Tech Compass 2025 Highlights Public Perception of AI and Its Workplace Impact Bosch’s Tech Compass 2025 report, unveiled at CES 2025, explores global attitudes toward AI and the growing demand for AI skills. Surveying 11,000 people across seven countries, the report highlights regional differences in AI acceptance, with 88% of respondents in China and 84% in India viewing technological development positively, compared to just 47% in France. The study reveals that one in four people has received workplace AI training, reflecting increasing adoption. Bosch has contributed to this trend, training 65,000 employees. The report emphasizes AI as a tool for standardizing tasks, illustrated by Bosch’s AI-powered HR assistant, ROB, which simplifies complex HR queries, and manufacturing AI systems that detect production failures. While AI is recognized as a transformative technology, Bosch acknowledges its current role as a productivity enhancer rather than a revolutionary game changer, with industries finding unique balances between automation and human interaction.

• Research on Agents: https://huyenchip.com//2025/01/07/agents.html Intelligent agents, as defined in AI research, operate by perceiving their environment and acting within it using sensors and actuators. The rise of foundation models has enabled the development of autonomous AI agents capable of performing complex tasks such as market research, data analysis, and business automation. These agents rely on tools for knowledge augmentation and capability extension, with planning and execution as critical components. Planning involves generating and validating action steps, while tools expand an agent’s functionality, from web browsing to executing code. Reflection and error correction are vital for improving performance, especially for tasks requiring multi-step execution. Efficient agent performance depends on well-designed tools, effective planning, and addressing failures in planning, tool use, and execution. These agents promise transformative productivity gains while posing challenges like cost, latency, and ensuring security.?

• Building generative AI applications is too hard, developers say | InfoWorld A recent IBM survey of 1,000 U.S. enterprise developers highlights the hurdles in generative AI development, including a significant skills gap, inadequate tools, and unclear processes. Only 24% of developers consider themselves generative AI experts, with challenges like the lack of standardized frameworks, difficulty in customization, and infrastructure complexity topping the list. Many rely on 5–15 tools, but performance, flexibility, and ease of use remain elusive traits. Despite these issues, AI coding assistants are widely used, saving up to two hours daily. The findings underscore the urgent need for intuitive tools and streamlined development stacks to support AI’s growing role in enterprise innovation.?

Trends?

The rise of low-code/no-code (LCNC) platforms, supercharged by generative AI, is democratizing technology by enabling non-technical users to create tailored solutions and fostering innovation across industries like BFSI and retail. Google, despite its immense resources, faces challenges with its Gemini AI model, needing a cohesive narrative, transformative use cases, and faster execution to reclaim its AI leadership. Sam Altman reflects on OpenAI’s groundbreaking journey and challenges in scaling safely, while financial pressures, including losses on the $200/month ChatGPT Pro plan, highlight tensions between innovation and profitability. These developments underscore the dynamic and competitive nature of the AI landscape, where strategic brilliance, adaptability, and a balance between ideals and business realities will determine long-term success.

• https://www.cdotrends.com/story/4364/unlocking-future-low-codeno-code-trends-and-predictions?refresh=auto The rise of low-code/no-code (LCNC) platforms, now supercharged by generative AI , is transforming application development by empowering non-technical users and accelerating digital innovation. This paradigm shift enables businesses to create tailored solutions faster, addressing challenges traditional methods often overlook due to cost or complexity. While adoption is surging globally, regions like North America and Southeast Asia are leading due to advanced digital literacy and innovation-driven cultures, whereas areas like MEA lag behind due to resistance to new methods and lack of citizen developer ecosystems. AI is the game-changer, reducing learning curves and enabling intuitive, natural language-driven development that democratizes app creation. GenAI’s integration into LCNC platforms is fueling faster innovation and cost-efficiency, making these tools indispensable for industries ranging from BFSI to retail. Emerging trends like conversational interfaces, multimodal user interactions, and modular UIs are reshaping user experiences. As platforms like Kissflow evolve by embedding AI deeply into their core, they highlight how GenAI is redefining workflows, driving agility, and fostering innovation in the tech landscape. The LCNC revolution, with GenAI at its heart, is poised to dominate 2025 and beyond, reshaping the software industry and democratizing technology.

• Google CEO Sundar Pichai emphasized urgency in advancing Google Gemini, admitting it lags behind rivals like ChatGPT but aiming to surpass them in 2025. While Meta invests heavily in AI and Apple focuses on embedding AI into its ecosystem, Google's nearly unlimited resources and infrastructure give it a chance to lead the AI race. However, Gemini's late start and weaker public perception put Google's trailblazer reputation at risk. Pichai underscored the importance of solving real user problems and scaling Gemini as a universal assistant to redefine AI standards in 2025. I guess we will have to see what happens.

My take: What can change this? First, Google needs a cohesive narrative for Gemini—one that’s compelling, differentiated, and directly addresses real-world problems. Right now, Gemini’s story isn’t resonating. Google must align its messaging with user priorities: practical use cases, seamless integration, and meaningful ROI. Second, reimagine the product launch strategy. Instead of chasing benchmarks that only tech insiders care about, demonstrate transformative applications for businesses and consumers. Show how Gemini can solve problems in healthcare, logistics, or education that no one else is tackling. Third, embrace partnerships to build ecosystems, not just products. OpenAI succeeded because it made ChatGPT ubiquitous by embedding it into tools like Microsoft Office. Google needs similar partnerships that make Gemini indispensable to enterprise workflows and everyday consumer tools.

Finally, speed up execution with relentless focus. Google has the talent and infrastructure, but its pace is glacial. Accelerate decision-making, simplify internal processes, and execute at startup speed. If Gemini is to avoid being just another "could have been" story, Google needs to shift from technical brilliance to strategic brilliance—and fast. The AI race isn’t about who has the most parameters; it’s about who wins hearts, minds, and wallets.rking harder is not going to help.?

• Reflections - Sam Altman? In a candid reflection, Sam Altman shared insights into OpenAI’s remarkable journey, from its founding nearly nine years ago to its pivotal role in shaping the future of AI. Altman detailed the challenges and triumphs of building OpenAI into a leader in artificial intelligence, including the surprise success of ChatGPT’s 2022 launch, which sparked unprecedented growth and catalyzed the AI revolution. Acknowledging the difficulties of scaling rapidly in uncharted waters, Altman emphasized the importance of governance, gratitude, and adaptability. He highlighted OpenAI’s evolving mission—from pioneering safe AGI to exploring superintelligence, which could accelerate scientific innovation and prosperity. Looking ahead, Altman believes 2025 will mark a milestone, with AI agents joining the workforce and reshaping industries. While the road has been tumultuous, Altman expressed confidence in OpenAI’s ability to iterate safely, adapt to societal needs, and guide AI development for broad benefit.

• Here is more on OpenAI is losing money on its pricey ChatGPT Pro plan, CEO Sam Altman says | TechCrunch OpenAI CEO Sam Altman revealed that the company is losing money on its $200-per-month ChatGPT Pro plan due to unexpectedly high usage. Initially launched as a premium offering, ChatGPT Pro provides access to advanced AI models and tools but has struggled with profitability. OpenAI, which reported a $5 billion loss on $3.7 billion revenue last year, faces mounting operational costs, including AI infrastructure and staffing. Altman acknowledged the lack of rigorous pricing studies for its plans and hinted at exploring usage-based pricing or increasing subscription fees to address financial challenges. Despite these hurdles, OpenAI projects ambitious growth, targeting $11.6 billion in revenue for 2025 and $100 billion by 2029, emphasizing the high stakes in scaling AI-driven services.

My take: Sam Altman’s reflections are heartfelt, but will his tune shift as OpenAI leans harder into profitability? Recent murmurs suggest cracks in the armor: Altman’s tweet about losing more money on the wildly popular $200/month Pro plan underscores the tension between innovation and financial sustainability. OpenAI, originally founded as a nonprofit with lofty goals of benefiting humanity, is now grappling with the realities of scaling a business. Here’s the hard truth: balancing world-changing ambitions with the demands of a for-profit model isn’t just difficult, it’s transformational. The success of Pro subscriptions shows people are willing to pay for value, but when profitability starts influencing strategy, compromises often follow. Will OpenAI continue to prioritize its mission of safety and empowerment, or will the bottom line dictate its trajectory? The next phase of OpenAI’s journey might reveal just how much sentiment and strategy diverge when profit margins come into play. One thing’s for sure: watching Altman navigate this crossroads will be a masterclass in balancing ideals with business realities.?

Concerns

Another week, another news about copyright violations or news about AI models trained on copyrighted material without permission … Anthropic has settled a dispute with music publishers over using copyrighted song lyrics to train its Claude AI, agreeing to strengthen guardrails and collaborate on addressing future concerns, while publishers push for clearer court rulings on fair use in AI training. Meanwhile, Meta faces allegations of knowingly using pirated books from the LibGen dataset to train its Llama models, with internal documents implicating CEO Mark Zuckerberg, sparking renewed legal challenges from authors. These cases highlight the unresolved tension between AI innovation and copyright law, signaling an urgent need for clearer frameworks and collaborative approaches to balance intellectual property rights with technological advancement.

• Anthropic reaches deal with music publishers over lyric dispute - The Verge The case centered on Claude AI allegedly using lyrics from over 500 protected songs during training without proper licensing. As part of the settlement, Anthropic agreed to maintain and enhance existing guardrails to prevent future infringements and collaborate with publishers to address any concerns. While Anthropic claims the use of copyrighted material for AI training falls under “fair use,” publishers continue to push for a court ruling to block the use of protected lyrics in future training efforts. This case highlights the growing tension between AI development and copyright law, signaling potential shifts in how intellectual property is handled in the era of generative AI.

My take: What happened with Anthropic marks a critical turning point in the AI and copyright conversation. Here’s the crux: generative AI is revolutionizing industries, but it’s also colliding head-on with intellectual property laws that were never designed to address these complexities. Anthropic’s settlement isn’t just a bandage, it’s a clear signal of how the legal framework is evolving in real time to meet the challenges posed by AI systems that thrive on vast datasets, often containing copyrighted material. By agreeing to maintain and strengthen guardrails, Anthropic is acknowledging a growing industry trend: AI companies need to proactively work with creators and rights holders to define boundaries, develop safeguards, and establish procedures for dispute resolution. The fact that Anthropic and the publishers reached an agreement signals a shift toward collaboration over litigation—a model that, frankly, more AI developers should consider.

This case also highlights the growing importance of establishing standards for responsible AI development, especially as the fair use doctrine remains a contested issue in the courts. Using copyrighted material for training AI models is still a legal gray area, and future rulings will play a critical role in defining its boundaries. Beyond song lyrics, this settlement underscores the broader challenge of balancing rapid AI innovation with intellectual property rights, pushing the industry towards creating more transparent and collaborative frameworks for the future.

• Meta knew it used pirated books to train AI, authors say | Reuters Ane one more copyright issue for you:? Meta is facing renewed accusations in a copyright lawsuit, with authors alleging the company knowingly used pirated books to train its AI models. Internal documents, disclosed during discovery, reportedly show that Meta CEO Mark Zuckerberg approved the use of the LibGen dataset, which contains millions of pirated works. The authors, including Ta-Nehisi Coates and Sarah Silverman, claim these actions were taken despite concerns raised by Meta’s AI executive team. The lawsuit centers on Meta's Llama language models and includes requests to revive dismissed claims about copyright infringement and metadata stripping. While Meta maintains that its use of copyrighted materials falls under fair use, the case highlights ongoing legal battles over the ethics of AI training datasets.?

Case Studies?

AI is advancing healthcare by transforming diagnostics and treatments for infectious diseases, while the Mayo Clinic uses NLP for earlier pancreatic cancer detection. The DHS Generative AI Playbook guides responsible public sector AI adoption. Retail sees AI-driven shopping trends, with consumer electronics dominating recommendations and privacy concerns persisting. Generative AI is reshaping law firms by streamlining tasks like drafting and e-discovery, requiring robust governance for successful adoption. Porsche leverages large language models like ChatGPT to enhance vehicle development efficiency through tailored AI tools. Beauty brands aim to unlock $9-10 billion in value by scaling generative AI for hyperpersonalization, rapid innovation, and improved consumer engagement.

Healthcare

• Challenges and applications of artificial intelligence in infectious diseases and antimicrobial resistance AI is transforming how we address infectious diseases and antimicrobial resistance (AMR), offering breakthroughs in diagnostics, treatment, and drug discovery. Machine learning models accelerate antibiotic discovery and enhance pathogen detection, resistance prediction, and personalized treatments. Challenges persist, including ethical concerns, data biases, and limited access to standardized datasets. AI-driven innovations, such as rapid bacterial phenotyping and the development of antimicrobial peptides, promise a faster response to antibiotic resistance. The field continues to evolve, emphasizing transparency, fairness, and collaboration among regulators, clinicians, and developers to address critical global health threats effectively.?

• Use of artificial intelligence tools and electronic health record data for pancreatic cancer risk prediction - Mayo Clinic The Mayo Clinic is leveraging AI and machine learning (ML) tools to enhance pancreatic cancer (PC) risk prediction using electronic health records (EHR). Recent studies by Dr. Shounak Majumder and colleagues developed natural language processing (NLP) algorithms to extract familial and genetic PC risk factors from unstructured clinical notes, achieving high sensitivity. These efforts aim to automate the identification of high-risk individuals and integrate this data into risk-based screening programs, addressing the significant challenge of diagnosing PC at earlier, more treatable stages. Future research will focus on enhancing these models with advanced tools like large language models, validating them in real-world settings, and identifying novel risk factors for sporadic PC within diverse population cohorts.?

Public Sector

• As we highlighted last week, DHS has been testing Gen AI for a while. They released a playbook along their finding: DHS Generative AI Public Sector Playbook | Homeland Security The Department of Homeland Security's Generative AI Public Sector Playbook outlines key lessons from pilot programs and provides actionable steps for responsibly adopting GenAI technologies. Designed for public sector organizations at any stage of AI adoption, the playbook covers technical, policy, and administrative considerations to guide effective implementation. Steps include resource assessment, internal alignment, and laying foundational groundwork for GenAI integration. Tailored recommendations address various organizational roles, from technical experts to policy personnel, ensuring comprehensive readiness for future AI developments. Direct download link: DHS Playbook for Public Sector Generative Artificial Intelligence Deployment??

Retail and E-commerce?

• Consumer Electronics Take Center Stage in AI-Driven Shopping Revolution A BrandRank.AI report reveals that consumer electronics dominate generative AI shopping searches, with 50% of users gravitating toward the category. Nearly half of AI users have made purchases based on AI recommendations, and 60% rely less on traditional search engines. Privacy concerns persist, with 63% worried about data transparency, but trust in GenAI platforms remains high (66%). Younger adults and higher-income groups lead adoption, while seniors show significant engagement in product research. The report also predicts CES 2025 trends, highlighting next-gen GPUs, advanced displays, and AI-powered home devices as key innovations.??

Legal

• Change Management: How to Finesse Law Firm Adoption of Generative AI – this article offers an excellent framework that can be integrated in any industry.? The integration of generative AI into law firms offers significant opportunities to improve efficiency, enhance service quality, and gain a competitive edge. However, successful adoption requires thoughtful change management to balance innovation with the legal profession's ethical and traditional practices. Andrew Ng’s AI Transformation Playbook provides a structured framework for managing this transition effectively. Pilot projects serve as the foundation for broader AI adoption, addressing skepticism by demonstrating tangible benefits in well-defined use cases. For example, AI can draft nondisclosure agreements or streamline e-discovery processes, allowing lawyers to focus on nuanced legal tasks. Early successes build trust and identify challenges, such as data management issues, before scaling. Creating a dedicated AI leadership team ensures alignment with firm goals while addressing ethical and operational concerns. This team includes partners, associates, IT professionals, and compliance officers who develop governance policies, training programs, and data security protocols. Training is critical to AI adoption, offering hands-on experiences to demystify the technology and guidelines to verify AI-generated outputs. A designated "review attorney" role ensures that all AI-driven work meets the firm’s rigorous standards, maintaining quality and trust. A coherent AI strategy aligns technology initiatives with firm goals, such as enhancing client satisfaction and operational efficiency. This strategy evolves with real-world feedback, regulatory changes, and advancements in AI capabilities, ensuring ongoing relevance and impact. Transparent communication fosters trust among staff and clients, addressing concerns and highlighting successes. Informal gatherings, newsletters, and town halls keep stakeholders informed, while external clarity reassures clients of ethical safeguards. Generative AI, when implemented thoughtfully, complements human expertise rather than replacing it. By handling repetitive tasks, it allows lawyers to focus on strategic endeavors, strengthening client relationships and advancing the firm’s reputation for excellence. This transformative approach positions law firms to thrive in a dynamic, AI-driven future.

Automotive

• AI in vehicle development: Large Language Models - Porsche Newsroom Porsche Engineering is leveraging large language models (LLMs), such as OpenAI's ChatGPT and Meta's LLaMa, to streamline and enhance efficiency in vehicle development. These AI-driven models, trained on proprietary engineering data, assist in revising customer specifications, automating the organization of technical requirements, and ensuring clarity in design documents. In a pilot project, LLMs reduced specification revision time by 50%, demonstrating the potential for substantial efficiency gains. The AI also supports error tracking during test drives by comparing new issues with historical data in real time, simplifying troubleshooting across vehicle platforms. Future applications aim to integrate AI into complex numeric data processing for faster insights, aligning with Porsche's strategy for data-driven innovation. Combining AI tools with human expertise allows Porsche to maintain engineering excellence while focusing on high-value tasks, setting a new benchmark in automotive development.

My take:? Porsche Engineering's approach to leveraging large language models (LLMs) like OpenAI’s ChatGPT and Meta’s LLaMa is exactly the kind of smart AI strategy I’ve been advocating for in telecom—and frankly, across industries. Why burn through cash reinventing the wheel when you can fine-tune what’s already out there?

The magic here is not in building an LLM from scratch (that’s a luxury only a handful of companies can afford); it’s in taking a commercially available model and training it on your proprietary data. Porsche’s move to teach these LLMs their engineering expertise, turning generic tools into tailored productivity engines, is a textbook example of how to innovate efficiently.

Here’s why it works:

1. LLMs are preloaded with broad knowledge: They’re like encyclopedias but not specialists. Training them with your data makes them masters of your domain.
2. Avoids cost bloat: Developing your own model means spending millions (or billions!) on data collection, training, and compute infrastructure. Porsche is spending that money where it counts—on its own engineering excellence.
3. Iterative improvement: LLMs aren’t static. Every project refines them further, making the tool smarter and better suited to future tasks.

It’s the same advice I’ve been giving to telecom: don’t waste precious resources creating your own AI models. Use what’s out there, train it on your network, operational, and customer data, and let iteration do the heavy lifting. This strategy scales, saves money, and keeps you competitive in a world where innovation waits for no one.

AI doesn’t need to be bespoke to drive business value—it just needs the right data and a clear strategy. Porsche gets it, and it’s time everyone else does, too.

Beauty

• How beauty players can scale gen AI in 2025 | McKinsey Beauty brands are poised to unlock significant potential with generative AI (gen AI), projected to contribute $9-10 billion to the global economy in the beauty sector alone. By focusing on high-impact use cases like hyperpersonalized targeting, experiential product discovery, rapid packaging-concept development, and innovative product creation, beauty players can improve consumer engagement and accelerate time-to-market. Strategic approaches include leveraging modular gen AI solutions tailored with company-specific data, implementing robust risk frameworks, and maintaining a human-in-the-loop for creative tasks. Scaling gen AI requires cross-functional collaboration, ongoing refinement through testing, and investments in organizational capabilities, positioning early adopters as leaders in this rapidly evolving space.??

Women Leading in AI?

Let’s walk down the history lane. Women in AI: From Pioneers to Modern Leaders Women have played an essential but often under-recognized role in shaping the field of Artificial Intelligence (AI). From the pioneering efforts of Ada Lovelace, the first computer programmer in the 19th century, to today’s leading innovators, their contributions have laid critical foundations and driven advancements across diverse areas of AI.

In the 20th century, Elaine Rich authored one of the earliest AI textbooks, influencing generations of AI professionals. Cynthia Breazeal pioneered social robotics, and Cynthia Dwork’s groundbreaking work in differential privacy revolutionized data security. In modern times, academics like Dr. Fei-Fei Li, Regina Barzilay, and Cynthia Rudin have advanced computer vision, natural language processing, and interpretable machine learning, respectively.

In the tech industry, leaders such as Joy Buolamwini (Algorithmic Justice League), Claire Delaunay (NVIDIA), and Rana el Kaliouby (Affectiva) have reshaped AI’s role in addressing algorithmic bias, robotics, and emotional AI. Public sector contributors like Marloes Pomp and Allie Miller have leveraged AI for social and public services, enhancing its reach and impact.

Tag a woman in AI who inspires YOU!?

Learning Center

Learning

• Best Generative AI (GenAI) Courses & Certificates [2025] | Coursera Learn Online? Discover Generative AI (GenAI) courses that focus on skills like model creation, natural language processing, and AI applications. Prepare for careers in AI development, data science, and machine learning engineering. You can select free courses based on your level! Happy learning.

Tools and Resources

• Gamma – a game changer in creative assets development, used to simplify the creation of visually engaging presentations, documents, and interactive content. It combines the flexibility of a slideshow with the depth of a document and the interactivity of modern web tools. Powered by intuitive design and customization features, Gamma enables users to craft professional, media-rich narratives quickly, making it ideal for pitches, reports, and storytelling. A free version is available, and then you can decide if you want to get a paid one.?

If you enjoyed this newsletter, please comment and share. If you would like to discuss a partnership, or invite me to speak at your company or event, please DM me.