GDSC Hult Weekly Update: June 4th - June 9th Newsletter
Databricks to Buy Data-Management Startup Tabular in Bid for AI Clients
Wall Street Journal (June 4, 2024)
In its second billion-dollar AI-related acquisition, following the $1.3 billion MosaicML deal last June, Databricks is now acquiring the data-management company Tabular. Built by Netflix veterans Ryan Blue, Dan Weeks, and Jason Reid, the startup provides cloud storage on top of Iceberg, an open-source table format that allows the storage of large amounts of data in data lakes. Databricks has not specified how much it paid for Tabular, but it is known that the amount was between $1-2 billion.
Like other competitors such as Snowflake, Databricks is aiming to harness one of the most valuable resources in the AI race: business data. Although these companies already provide platforms for storage, organization, and analysis of data across cloud providers, which will allow for further AI model customization, including Retrieval-augmented Generation(RAG), a way of customizing AI models with business data for analysis, there is a need to ease the usage of different data formats in the cloud without increasing costs. This incentive is crucial for customers to produce and store more data.
According to CEO Ali Ghodsi, projects like Delta Lake UniForm, enabling data engineers to use multiple open-source table formats simultaneously, are already in the pipeline. These projects, led by Iceberg’s creators, aim to enhance compatibility. However, Ramke Ramakrishnan, an analyst at Gartner, notes the risk that the formats may never converge, allowing major vendors like Microsoft to develop their own multi-format tools.
With over $1.6 billion in revenue at the end of the fiscal year on January 31, a growth rate of 50% from the year before, and a valuation of $43 billion, Databricks is observing signals such as cloud software listings in the IPO market before going public, according to Ghodsi.
Nvidia overtakes Apple as its market capitalisation surpasses $3tn
Financial Times (June 6, 2024)
As big tech groups such as Google, Microsoft, and Meta increase their expenditure on Nvidia’s chips, investor confidence in the chipmaker has risen, boosting its share price. Last week, Nvidia’s market valuation reached $3.01 trillion, surpassing Apple’s for the first time and making it the second most valuable company in the world. According to chief executive Jensen Huang, Nvidia is driving a new “industrial revolution” by enhancing global business productivity with its AI models, contributing to more than a third of the gains on Wall Street’s benchmark S&P 500 index this year.
Nvidia reported a 262% increase in revenues, driven largely by its “Hopper” chips, demonstrating its unique ability to generate AI revenues, said Stuart Kaiser, head of US equity trading strategy at Citigroup. Despite competitors like AMD and Intel vying for market share, Nvidia remains the leading provider in the global technology race, supplying both hardware and software to meet growing AI demands.
Looking ahead, Huang has promised a “one-year rhythm” of new chips, unveiling “Blackwell” products in March and teasing the upcoming “Rubin” processors at Taiwan’s Computex conference. Apple’s exclusion from the AI market and declining iPhone sales due to competition in China could be mitigated by plans to integrate AI features into its products, discussed at their Worldwide Developers Conference on June 10. CEO Tim Cook is optimistic about Apple’s future in AI, and the company's shares are performing well, bolstered by a $110 billion share buyback announced in May.
As for Nvidia, the company noted that its valuation of 42 times expected earnings for the next 12 months is not at an all-time high when measured against its historic or projected profits, indicating a complex future ahead.
Google looks to AI to help save the coral reefs
TechCrunch (June 6, 2024)
Calling in Our Corals is a Google project where the public listens to over 400 hours of reef sounds from around the world and clicks whenever they hear a fish sound. This data trains SurfPerch, an AI tool developed with Google Research and DeepMind. SurfPerch enables scientists to monitor reef health, track nocturnal activity, and study reefs in deep or murky waters without needing GPU processors. This approach opens new opportunities for understanding reef communities and aiding conservation efforts.
Researchers discovered that bird sounds, despite being different, share patterns with fish sounds and can also train the model. Combining data from Calling in Our Corals with SurfPerch in initial trials revealed differences between protected and unprotected reefs in the Philippines, tracked restoration outcomes in Indonesia, and provided insights into fish community dynamics on the Great Barrier Reef. The project continues today, with new audio added to the website to further train the AI model, according to Google.
Germany’s Merck Bets on AI Drug-Design Partnerships, Rules Out Acquisitions – Interview
Wall Street Journal (June 7, 2024)
The pharmaceutical company Merck, known as EMD Group in the U.S., is following the AI trend in drug testing with a different approach. The company seeks to partner with AI firms rather than acquiring them, a strategy CEO Peter Guenter considers more prudent. “Acquiring such a company is probably not a good idea because the field is moving so fast that we have to leave those companies in their own ecosystem so they can continue to evolve,” he explained.
Similarly, Merck has decided to avoid the weight-loss drug market, despite the popularity of Novo Nordisk’s Ozempic and Wegovy and Eli Lilly’s Mounjaro and Zepbound. The company is focusing on its core therapeutic areas of oncology, neurology, and immunology. This cautious approach reflects Merck's strategy to concentrate on areas where it has the strongest expertise and potential for impact.
Merck's hopes for AI are currently in the initial research phase, a long, risky, and costly process. While AI has been used in drug development, Guenter believes its biggest impact will be in the research phase, potentially increasing the success rate of molecules in clinical development. A 2022 Morgan Stanley report predicted that 50 new AI-developed drugs could enter the market globally in the next decade.
Merck recently entered AI drug-development collaborations with American-Israeli biotech company Biolojic Design and U.S.-based Caris Life Sciences. Last September, the company partnered with U.K.-based AI specialists BenevolentAI and Exscientia, aiming to significantly reduce drug discovery timelines. These collaborations are expected to accelerate the process by 50-60%. While uncertainties persist and humans remain the highest authority, Guenter anticipates some savings and, more importantly, shorter drug development and testing timelines.
Deal Dive: Human Native AI is building the marketplace for AI training licensing deals
TechCrunch (Juna 8, 2024)
London-based Human Native AI is creating a marketplace to connect companies building LLM projects with individuals willing to license their data. AI companies need vast amounts of data to train their models, and data holders should be compensated for making their data available. The startup offers a platform for these transactions at no cost to rights holders, helping them prepare and price their content. Human Native AI makes money by charging AI companies for transactions and monitoring services.
James Smith, CEO and co-founder of Human Native AI and a former Google DeepMind employee, pitched his idea to his friend Jack Galilee, an engineer at GRAIL. They observed that many AI companies, including DeepMind, struggled with acquiring high-quality data. With the European Union AI Act and potential AI regulation in the U.S., sourcing data ethically will become increasingly important. Two months into operations and currently in beta, the startup has seen strong demand from both sides. This week, they announced a £2.8 million seed round led by LocalGlobe and Mercuri, two British micro VCs, to support team building.
Smith also believes that smaller AI systems, which lack the resources to secure deals with major content providers like Vox or The Atlantic, should have access to quality data. “One of the major challenges with licensing content is large upfront costs, which restrict who you can work with,” Smith said. “How do we increase the number of buyers for your content and reduce the barriers to entry? We think that is really exciting.” He mentioned the future benefits for rights holders, who will be able to build their prices based on data deal history on the platform.
Despite their enthusiasm, Smith emphasized the responsibility to support the industries that have brought them to this point. “We are AI optimists on the side of humans,” he said, stressing the importance of finding ways for everyone to participate.
Adobe uproar shows no one trusts Big Tech
Morning Brew (June 9, 2024)
Adobe’s recent announcement sparked user outrage, causing many to reconsider their trust in the company. The updated terms of service stated, “we may access your content through both automated and manual methods.” Screenshots of this update went viral on X, leading users like TV personality Alton Brown to suspend use of Adobe products. The concern was that Adobe’s generative AI model, Firefly, might be trained on user data.
Adobe responded, assuring that user data won't be used to train its generative AI models, noting that the TOS language has remained largely unchanged for over a decade, as highlighted by The Verge and The Register. Chief Product Officer Scott Belsky admitted the wording was “unclear,” emphasizing the importance of trust and transparency. This incident highlights the need for companies like Adobe, already scrutinized for its dominance in creative software, to handle user data carefully amidst rising concerns over unauthorized use of material.
Credits:
- Rafaella Andrade
- GANGA SINGH
- GDSC Events Team
References
- Morning Brew. (2024, June 9). Adobe uproar shows no one trusts big tech. Morning Brew. Retrieved from https://www.morningbrew.com/daily/stories/2024/06/09/adobe-uproar-shows-no-one-trusts-big-tech
- Szkutak, R.(2024, June 8). Deal Dive: Human Native AI is building the marketplace for AI training licensing deals. Retrieved from https://techcrunch.com/2024/06/08/deal-dive-human-native-ai-is-building-the-marketplace-for-ai-training-licensing-deals/#:~:text=Human Native AI is a,opt in and are compensated.
- Smolack, H. (2024, June 7). Germany’s Merck Bets on AI Drug-Design Partnerships, Rules Out Acquisitions – Interview. Retrieved from https://www.wsj.com/tech/biotech/germanys-merck-bets-on-ai-drug-design-partnerships-rules-out-acquisitions-interview-943c5b7a#
- Lin, B. (2024, June 4). Databricks to buy data-management startup Tabular in bid for AI clients. The Wall Street Journal. Retrieved from https://www.wsj.com/articles/databricks-to-buy-data-management-startup-tabular-in-bid-for-ai-clients-829e5bcf
- Acton, M. Megaw, N. (2024, June 6). Nvidia overtakes Apple as its market capitalisation surpasses $3tn. Retrieved from https://www.ft.com/content/f5f7bce2-6a97-42a0-80ca-74468cf8c023
- Perez, S. (2024, June 6). Google looks to AI to help save the coral reefs. TechCrunch. Retrieved from https://techcrunch.com/2024/06/06/google-looks-to-ai-to-help-save-the-coral-reefs/