20 tech trends to watch closely in 2024

Makesmartertech decisions. Our platform will help youdiscover,understand, and manageyour response toemerging tech. Make confident tech decisionsthat drive growth, improve operations, andbeat the competition with CB Insights. ▮▮▮▮IIIITry our platform–free!Sign up for a free trial Your alwaysonAI-DrivenResearch Analyst Do less searching, get more answers ▮▮▮▮IIIIJoin the waitlist forearlyaccess Tech trends AI Healthcare&life sciences GPU shortage forces companies to be smarter05Multimodal AI will rise on corporates’ wish lists12Synthetic data bonanza17 The great AI drug race heats up71Digital therapeutics (DTx) & wellness consolidates78The brain becomes a fierce tech battleground84 Enterprise Retail&consumer DIY software development upends engineering orgs22Global employment & payroll market shakes out28Quantum computing advances hint at faster commercialization33 AI sales agents flood the e-commerce landscape89Retailers tackle shrink with AI loss prevention96AI sends a shockwave through gaming102 Cybersecurity Industrials Cyber chaos drives security consolidation38AI vs. AIdogfightsredefinedata security46 Humanoid robots come for manufacturing109 Venture Financialservices&insurance Corporate venture refocuses on strategic fit115Unicorns need a new playbook to survive120 Banks get AI FOMO52Blockchain’s uphillfinservbattle59Extreme weather is an opportunity forinsurtech66 AI GPU shortageforces companiesto be smarter AI Companies are desperate for GPUs H100ssell at a massive mark-up “The GPUs at this pointare considerably harderto get than drugs.” Elon Musk at the WSJCEO Council Summit AI Startups link up with big tech to access their AI chipsand compute power GenerativeAI companies with two or more big tech investors(as of 1/30/2024) AI While also looking outside of big tech for cheaper cloud compute Nvidiais giving priority access to its GPUs to smaller specialty cloudprovidersCoreWeave and Lambda Labs,which are offering cloud GPU access atlowercosts to enterprises AI But as demand continues to outstrip supply, companies aimto optimize their AI models to reduce compute time “For the deployment side of things, we found that the performance of our training processwas quite slow, especially when it gets into these large language models and when youtrain from scratch.MosaicMLoffers what's called programmatic optimization, which isnot so much on the hardware side of things, but rather on the algorithmic side. Can youfind ways of optimizing the time it takes to get to a certain performance bar?... We didexperiments where we usedMosaicMLprogrammatic optimization versus not, and we didsee a12% compute hour reduction for some of the models.” MosaicMLcustomerSenior Manager, Data Science at $1B+ valuation technology company AI Startups building software to help AI models run efficiently onavailable hardware gain traction AI GPU shortage also opens the door for experimenting with novelprocessor approaches AI Multimodal AI willrise on corporates’wish lists AI Multimodal AI is in its infancy, but models are quickly evolving Googleintroduces first commercial model taking diverse inputs in December 2023 AI More LLM developers will follow in Google’s footsteps to remaincompetitive Limited modalitiescommercially available In the roadmap or notannounced AI Models that can accept diverse inputs open up new opportunitiesfor AI in industries that are highly multimodal Googlewill bring Gemini-based models into its MedLM family of foundationmodelsfine-tuned for healthcare use cases in coming months “Medicine is an inherently multimodal discipline.When providing care, clinicians routinelyinterpret data from a wide range of modalities including medical images, clinical notes, lab tests,electronic health records, genomics, and more. Over the last decade or so, AI systems haveachieved expert-level performance on specific taskswithin specificmodalities…But how do webring these capabilities together to build medical AI systems that can leverage informationfrom all these sources?” GregCorrado, Head of Health AI, Google ResearchYossi Matias, VP, Engineering and Research, Google Research AI Watch for commercial applications across healthcare, automotive,retail, and manufacturing AI Synthetic data bonanza AI We are running out of high-quality data to train AI models Researchers estimate that, by2026,we will exhaust high-quality text data*for trainingLLMs—a trend that can slow downAI progress. AI Scraping proprietary sources is getting harder “Scraping proprietary data to train generative AI models for commercialapplication is nowuniversally recognized as an unsustainable business tactic.It's fraught with legal, financial, and reputational challenges.“ Shutterstock CEO Paul Hennessy, Q4’23 earnings call Source: CB Insights—Earnings Transcripts Analytics(date reflects quarter call occurred); media reports. AI Cost and scarcity drive model developers to experiment with synthe