Key takeaways

• The European Parliament approved the much-awaited AI Act in March, which aims to prevent the misuse of high-risk AI while promoting innovation. It is expected to be enacted in May/June 2024, and member states are responsible for establishing their own national oversight agencies.

• The GenAI ecosystem raised USD 6.5 billion (grew 59.0% QoQ) across 81 funding rounds, representing the highest funding amount within a quarter since 2021.
• Three major rounds—Amazon’s USD 2.75 billion investment in Anthropic, Chinese chatbot developer Moonshot AI’s USD 1 billion raise, and Infrastructure provider Kyndryl’s USD 500 million post-IPO debt—collectively accounted for 65% of the total funds raised during the quarter.

• Foundation Models: OpenAI launched its new model Sora in text-to-video generation, while Stability AI introduced Stable Video 3D. Similarly, Google expanded its portfolio with two video models. Most other updates were new model introductions, with startups like Anthropic (Claude 3), Mistral AI (Mistral Large), and Inflection AI (Inflection-2.5) launching models to compete with OpenAI’s GPT-4.

• GenAI Infrastructure: Incumbents exhibited significant activity in the development of GPUs and AI chips, with NVIDIA launching Blackwell GPU and Meta unveiling a new GPU cluster infrastructure, while Qualcomm released the Snapdragon 8S Gen 3 chipset, which would help to handle large models like those launched by Mistral AI and provide better performance. The emphasis on on-device processing too saw significant growth, with both incumbents and disruptors introducing new solutions, including RTX 500 and 1000 GPUs by NVIDIA and those by startups like DEEPX and Expedera.

• GenAI Applications: Incumbents and disruptors continued to add new features to enhance existing products, including voice capabilities to chatbots (e.g., Character.ai, OpenAI) and the ability to add sound effects to videos (e.g., Pika) as well as updating underlying models and other services to offer more precise and personalized replies for chatbots (e.g., Google Gemini, GPT Store).

• Multiple partnerships focused on integrating FMs into mobile phones were announced, including Google’s partnerships with MediaTek and Samsung. Developers continued broadening the accessibility of their models on cloud platforms, with Mistral AI leading the way by collaborating with major cloud providers such as IBM, Amazon, and Microsoft. 

• NVIDIA partnered with Lenovo, Cohesity, Amazon, Microsoft, Cisco, and Oracle to offer integrated data security, management, and computing infrastructure solutions. Disruptors Cerebras Systems, VAST Data, and Lambda reported similar partnerships. 

• Collaborations were also reported toward incorporating AI capabilities directly into hardware devices, including OpenAI releasing a ChatGPT app for Apple's Vision Pro headset and Adobe making its text-to-image tool available in the Apple Vision Pro headset.

• More than 10 M&A deals were reported within the quarter. Infrastructure provider Databricks dominated, with two new acquisitions: Einblick and Lilac. We tracked two acquisitions in the GenAI Applications space, with marketing content generator Jasper acquiring Clipdrop and Typeface acquiring TensorTour. Notably, Apple acquired DarwinAI and Brighter AI ahead of a major expansion into GenAI across its products and services planned for this year.

• Rising prevalence of small language models (SLMs) - SLMs present a potential solution to the drawbacks associated with large language models (LLMs), as they are designed to be more streamlined, requiring fewer parameters and less training data. This makes them faster and more cost-effective to train, as well as easier to deploy, particularly on smaller devices or in settings with limited computing power. Additionally, the capability of SLMs to undergo fine-tuning for particular applications offers increased adaptability and customization. Microsoft has reportedly been developing smaller and more economical GenAI models, supplementing its Phi-2 model launched in Q4 2023. Google also introduced Gemma, a model compact enough to run on a personal computer. As small and medium-scale enterprises increasingly embrace GenAI technology, demand for SLMs is poised to rise, driven by their cost-effectiveness and minimal resource requirements.

• GenAI’s expansion by integrating with hardware devices - As capabilities evolve, GenAI application developers are exploring ways to broaden GenAI applications by integrating GenAI directly into hardware. Noteworthy developments included OpenAI launching the ChatGPT app and Adobe making its text-to-image tool compatible with Apple's Vision Pro headset. Additionally, Brilliant Labs introduced lightweight AR glasses, with AI assistant Noa, while reports suggest that Google may incorporate Gemini into its earbuds. These integrations hold significant implications for the future of digital content creation and user experience, enhancing creativity and interaction in virtual environments.

• Advancement in open-source models - While open-source models have been in development for some time, initial studies indicated that their performance lagged behind proprietary models. However, recent advancements have shown a significant improvement in the quality and performance of open-source models, particularly with leading companies like Mistral AI and Stability AI introducing models with performance closer to their proprietary counterparts. Meta's open-source artificial general intelligence (AGI) initiative and xAI’s decision to open-source the base code of the Grok AI model on GitHub demonstrate a shift toward collaboration and transparency in the development of AI technologies. This should enable new features, unique applications, and design interfaces, as developers can leverage open-source models without having to build from scratch.

Regulations: Global benchmark set by EU with landmark AI Act

Analyst Take: The highlight of this quarter was the EU AI Act. This can be deemed a proactive step toward AI governance, balancing innovation and ethics. It will likely be a global benchmark for AI regulation and promote ethical AI practices. The Act will come into force by mid-2024, with enforcement over 2025–2026. For non-compliance, fines as high as USD 38 million or 7% of a company's global revenue will be imposed.

• European Parliament approves AI Act: The European Parliament approved the AI Act in March, which aims to prevent the misuse of high-risk AI while promoting innovation. The legislation will regulate FMs and GenAI, which are trained on large volumes of data. The law requires companies to publish detailed summaries of the content used for model training and perform model evaluations. The regulation is anticipated to take effect in May/June, following the completion of final checks and approval from the European Council. National authorities will enforce these regulations, with support from the AI office within the European Commission. Member states are responsible for establishing their own national oversight agencies. With this, citizens can lodge complaints regarding AI systems and obtain explanations concerning decisions. 

• California Bill proposes mandatory AI testing before release: California lawmakers proposed a bill to make companies test AI models before release. The proposed rules include mandatory testing of new AI models for “unsafe” behaviors and developing safeguards against hacking. Additionally, the rules require AI to be developed in a way that would allow it to be halted entirely if necessary. This aims to regulate AI technology and potentially set a precedent for similar rules across the rest of the nation.

Source: Compiled by SPEEDA Edge based on various sources

Other notable industry updates

• In January, the Federal Trade Commission launched an inquiry into investments and partnerships of major AI companies, including Amazon, Alphabet, Microsoft, Anthropic, and OpenAI, focusing on their investments and partnerships with cloud service providers. Notably, the European Commission also initiated an investigation into Microsoft's investments and partnerships with OpenAI and Mistral AI, amid competitive concerns. 

• The World Health Organisation introduced guidance on AI ethics and governance of large multi-modal models (LMMs). The guidelines consist of over 40 recommendations for governments, tech companies, and healthcare providers to ensure the proper usage of LMMs to protect public health and for developers to account for risks such as false, inaccurate, biased, or incomplete statements being produced. 

• European Commission to establish AI factories to provide easier access to dedicated supercomputers for startups in Europe, aiming to foster the development of general-purpose AI models. This initiative follows EU regulations on large AI models and seeks to support AI startups and researchers by offering resources (algorithm development, testing, and validation) and access to AI-dedicated supercomputers.

• US cloud companies to be compelled to disclose foreign AI developer customer data by the US Commerce Secretary to identify instances when foreign entities use cloud services to train LLMs for AI applications, which can hold potential for malicious cyber activities. This aims to tighten regulations on cloud services, given fears that Chinese companies could exploit cloud platforms to circumvent semiconductor restrictions.

• Google DeepMind creates an AI safety and alignment unit to address concerns over the potential misuse of its GenAI model Gemini. The new unit will be responsible for preventing AI models from giving out faulty medical advice, safeguarding children's interests, and ending the promotion of biases and injustices. This initiative follows the launch of the AI Alliance by IBM and Meta in the previous quarter, highlighting a collective effort toward self-regulation within the AI community.

Funding: Q1 2024 sees record funding, signaling strong investor confidence

Analyst Take: During Q1 2024, the GenAI ecosystem grew substantially, securing USD 6.5 billion in funding. This was a 59.0% increase from the previous quarter and represents the highest funding raised within a quarter since 2021. Major funding rounds from industry giants like Amazon/Anthropic and startups like Moonshot AI highlighted the willingness of investors to commit significant resources to promising AI ventures. Seed and early-stage funding rounds accounting for 84.0% suggests a healthy pipeline of emerging startups and ongoing innovation, which should help to sustain the funding momentum for the GenAI sector.

GenAI ecosystem Q1 2024 funding summary

• In Q1 2024, companies in the GenAI ecosystem raised USD 6.5 billion in 81 rounds. This marks a substantial increase of 59.0% compared with the previous quarter, representing the highest funding raised in a quarter since 2021. All three hubs contributed to this growth, with the highest having come from the GenAI Applications hub (up 137.2% QoQ).

• The growth in funding was supported by three deals: Amazon’s USD 2.75 billion investment in Anthropic, Chinese chatbot developer Moonshot AI's USD 1 billion raise, and Infrastructure provider Kyndryl’s USD 500 million post-IPO debt, which collectively accounted for 65% of the total funds raised during the quarter. 

• 19 companies in the business process improvement segment of the GenAI Applications hub raised USD 633 million, indicating a doubling of the amount of funding raised QoQ (contributing to a third and 41.3% of the GenAI Applications hub’s funding amount and number of rounds, respectively). This increase underscores the growing recognition among businesses across various industries of GenAI’s value in enhancing their processes.

• The average deal size in Q1 2024 rose 67.0% QoQ, with the most notable increase coming from GenAI Applications at 1.4x QoQ, while the average deal size of FMs also doubled. Notably, the average deal sizes for both hubs reached their highest points.

• Seed and early-stage funding rounds comprised 84.0% of total funding rounds during the quarter and Q4 2023; this is in line with the average for 2023, 81.7%. Although the early-stage funding rounds increased by 59.1% during the quarter, this was offset by a 35.3% decline in seed funding rounds. This suggests a possible progression in the industry beyond its initial stages of development.

Notable funding rounds and investors

• The top 10 funding rounds for Q1 2024 had a combined value of USD 5.3 billion, accounting for more than 80% of total funds raised within the quarter. Amazon’s investment in Anthropic (USD 2.75 billion) took the lead, while Chinese chatbot developer Moonshot AI was second with USD 1 billion. GenAI Infrastructure companies Kyndryl and Lambda raised USD 800 million in total. Amazon, Alibaba Group, and US Innovative Technology were notable investors in these rounds. 

Repeat funding rounds during TTM Q1 2024

Analyst Take: Repeat rounds, such as in the case of Anthropic, indicate sustained investor interest and a push to strengthen expertise and product deployment in GenAI during the past 12 months. 

• Anthropic was the only company to record five funding rounds during the trailing 12 months (TTM) to Q1 2024 (USD 5.0 billion), while Coreweave reported four (USD 3.4 billion). 
• Four companies, Cranium, Hippocratic AI, Mistral AI, and Together AI, reported three funding rounds each, while 40 companies reported two funding rounds each.

Repeat funding rounds during TTM Q1 2024

Product updates: Rivalry among model developers expands into the chatbot space

Foundation Models

Analyst Take: Several new models aimed at competing with OpenAI's GPT-4 were launched this quarter. Mistral AI introduced Mistral Large, while Chinese player Baichuan unveiled Baichuan 3, claiming superiority in processing tasks in the Chinese language. Inflection AI launched Inflection 2.5, boasting efficiency, using only 40% of OpenAI's computing power. Google's new model, Google Ultra, purportedly edged past GPT-4, while Anthropic's Claude 3 Opus model claimed superiority over GPT-4 and Google Ultra. Despite these, OpenAI remains determined, reportedly gearing up for a mid-year release of its next model, GPT-5. OpenAI has also reduced API access pricing and introduced new model versions, demonstrating its commitment to maintaining market dominance.

• We tracked 47 product updates in Q1 2024 among 16 disruptors (27 product updates) and eight incumbents (20 product updates).

• Most were new model introductions; startups like Cohere (Command-R), Anthropic (Claude 3), Mistral AI (Mistral Large), Inflection AI (Inflection-2.5), Reka (Reka Flash), and Zhipu AI (GLM-4), as well as incumbents like Google (Gemini 1.5) and SenseTime (SenseNova 4.0), continued to introduce upgrades to existing models. Furthermore, Cohere (Aya), Apple (MLLM), and Meta (Code Llama 70B) launched open-source models.
• OpenAI launched Sora, entering the text-to-video generation space. Stability AI introduced Stable Video 3D, whereas Google expanded its portfolio with two video models, VideoPoet and Lumiere, enriching the video-generation technology landscape.

• Incumbents broadened the application of GenAI during the quarter by incorporating their models into various business sectors. Google incorporated GenAI into Maps, Google Ads, and Chrome, enhancing user experiences across these platforms. Meanwhile, Amazon rolled out the Titan Image Generator to produce images on Fire TV devices, showcasing the versatility of GenAI in enriching multimedia content, and Meta announced plans to add GenAI to its Ray-Ban smart glasses. 

GenAI Infrastructure

Analyst Take: While maintaining its market dominance, NVIDIA introduced several chips to compete with industry rivals AMD and Intel. The standout among them is the Blackwell B200 GPU touted as the “world’s most powerful chip” for AI. Concurrently, Meta joined the ranks of customers venturing into chip development, following in the footsteps of Microsoft and Google, who launched their own chips last quarter, intensifying competition. With increasing demand for GenAI applications, AI chips are poised to become the next major arena for competition in the race for AI dominance. 

• We tracked 52 product updates in Q1 2024 among 24 disruptors (24 product updates) and 16 incumbents (28 product updates).

• Incumbents exhibited significant activity in developing GPUs and AI chips, with several notable launches to enhance performance. NVIDIA has been particularly active, introducing a range of new products including the advanced Blackwell GPU and the RTX 2000 Ada generation GPU. Meta unveiled a new GPU cluster infrastructure, and Qualcomm released the Snapdragon 8S Gen 3 chipset, while AMD launched the Ryzen 7 8700G. These GPUs and AI Chips are designed to handle the larger models like those launched by Mistral AI or provide better performance, speeds, or functionalities to cater to the expanding AI landscape.

• On-device processing witnessed significant growth with both incumbents and disruptors introducing new solutions. NVIDIA unveiled its RTX 500 and 1000 GPUs specifically designed for laptops with compact designs and three desktop graphics chips to improve AI functionalities in PCs. Ambarella introduced the N1 SoC series GPU solution, while Qualcomm launched a model deployment solution to advance large model processing capabilities. Disruptors like DEEPX and Expedera also entered this space by launching new hardware products to bolster on-device processing, while Syntiant Corp launched model deployment solutions to optimize computing capability. All these products would help to expand the scope of on-device processing.

• In response to growing concerns about GenAI model security, there were product launches from CalypsoAI (CalypsoAI Platform), Microsoft (Copilot for Security), Cloudflare (Firewall for AI), Akto (GenAI Security Testing), iboss (ChatGPT Risk Module), Metomic (ChatGPT browser plugin), and Dynatrace (AI observability tool), which should help enhance the integrity of AI systems to some extent.

GenAI Applications

Analyst Take: GenAI chatbots have evolved to become essential tools across various industries, with advancements in multimodal systems, where these chatbots can comprehend language, visuals, and voice inputs. The focus is now shifting toward enhancing the accuracy and personalization of responses. For instance, Google upgraded its chatbot with a more robust underlying model, while OpenAI introduced a memory feature and a GPT Store to facilitate the creation of customized bots.

• We tracked 41 product updates during Q1 2024 across 15 disruptors (23 product updates) and six incumbents (18 product updates). 

• Both disruptors and incumbents continued to add new features to enhance existing products. Some notable new capabilities are as follows:
  • Adding voice capabilities to chatbots (e.g., Character.ai, OpenAI)
  • Updating underlying models and other services to offer more precise replies for chatbots (e.g., Google Gemini, You.com)
  • Offering a more personalized experience (e.g., OpenAI’s launch of GPT Store and memory feature)
  • Adding sound effects to videos (e.g., Pika)

• Several FM developers also entered the chatbot space; Mistral AI launched a new chat assistant “Le chat,” Krutrim launched a chatbot called Krutrim AI, Hugging Face introduced free customizable GenAI chatbots named Hugging Chat Assistants, and NVIDIA launched Chat with RTX for personalized responses using localized datasets.

• Meta announced plans to label and identify AI-generated images on its platforms, such as Facebook and Instagram, while OpenAI announced plans to add watermarks to images created by its image generator DALL-E 3. This should somewhat address concerns about the transparency of AI-generated content.

Partnerships: Momentum grows in hardware integrations and chip development

Analyst Take: A trend of incorporating FMs into smartphones seems to be gaining momentum. Samsung, for instance, partnered with Google Cloud to incorporate models available on Vertex AI cloud, such as Gemini Pro and Imagen 2, offering a range of capabilities from creating summaries to photo editing. Additionally, the Gemini Nano model will support Google Messages' "Magic Compose" feature, enabling the drafting of messages even without internet connectivity through on-device processing. While most partnerships reported during the quarter focused on cloud processing, Google also collaborated with MediaTek to use the Gemini Nano model in developing chips to enable AI features on mid-range phones. This potentially signals a shift toward integrating FMs directly into AI chips, underscored by the emergence of SLMs that can run chips.

• We observed 82 partnerships in Q1 2024, with disruptors accounting for more than 70% of the companies involved. Most GenAI partnerships reported were in the FMs hub (46%), and product partnerships (73%) were the most common. Product partnerships accounted for ~90% of the GenAI Applications and GenAI Infrastructure hubs, but only 53% in FMs (customer partnerships and sales partnerships in the FMs hub were 26% and 21%, respectively).

• Multiple partnerships focused on integrating FMs into mobile phones were announced. Google collaborated with MediaTek to leverage Gemini Nano for developing chips, enabling AI capabilities on mid-range phones. Additionally, Google Cloud partnered with Samsung to enhance Galaxy S24 phones with Gemini AI, while discussions were underway for Apple to integrate Gemini AI and Baidu’s models into iPhone software. Chinese tech giant Baidu also teamed up with Samsung and Lenovo to incorporate GenAI into their smartphones.

• Developers continued to broaden the accessibility of their models on cloud platforms. Mistral AI led the way by collaborating with major cloud providers like IBM, Amazon, Snowflake, and Microsoft to make its models available on the cloud. Stability AI, Hugging Face, and Deci AI also partnered with other players to introduce advanced GenAI models. 

• Building on previous partnerships with news companies, OpenAI sustained its progress by collaborating with Le Monde and Prisa Media to access news content for model training. Discussions were underway with CNN, Fox, and Time to secure content licenses for AI development. Google also partnered with Reddit to use its data API for training AI models.

• NVIDIA partnered with Lenovo, Cohesity, Amazon, Microsoft, Cisco, and Oracle to offer integrated data security, management, and computing infrastructure solutions. Several disruptors including Cerebras Systems (partnered with Qualcomm), VAST Data (partnered with Run:AI), and Lambda (partnered with Quantiphi) also reported similar partnerships.

• Partnerships were also aimed at providing integrated solutions to aid developers in streamlining their app development process, thus offering time and cost savings. This includes Stack Overflow partnering with Google Cloud to enhance AI capabilities for developers, Anaconda partnering with IBM watsonx to deliver enterprise-scale AI solutions, Airbyte partnering with Vectara to streamline GenAI application development, and Siemens and AWS partnering to facilitate the use of GenAI in software development.

• Another notable trend during the quarter was strategic collaborations to incorporate AI capabilities directly into hardware devices, improving user experiences and functionality.
  • OpenAI released a ChatGPT app for Apple's Vision Pro headset.
  • Adobe made its text-to-image tool available in the Apple Vision Pro headset.
  • Perplexity partnered with Rabbit to integrate conversational AI into its R1 device.
  • Microsoft partnered with Samsung to introduce intelligent connectivity features in the Galaxy Book4 Series.
  • OPPO partnered with Microsoft to bring Microsoft Copilot to its smartphones.

• Notable collaborations to deliver integrated products within the marketing content space include Bounteous x Accolite partnership with Persado to advance commerce and marketing experiences for customers and Typeface’s integration with Microsoft Dynamics 365 Customer Insights for marketing campaign management.

Incumbent engagement in the startup ecosystem

• In Q1 2024, incumbents put in significant funding into startups, led by Amazon fulfilling its USD 4 billion commitment made in September to Anthropic. Microsoft followed suit by investing in Mistral AI, akin to its existing partnership with OpenAI. Chinese tech giant Alibaba was a lead investor in both Moonshot AI and 01.AI funding rounds. Additionally, NVIDIA invested in Abridge and Kore.ai, Meta and Apple participated in the SuperAGI funding round, KT Corp participated in the Rebellions funding round, and Google co-led the Send AI funding round, further indicating the growing interest and support from established players in the startup ecosystem. 

• NVIDIA led the incumbent collaboration with disruptors in Q1 2024, partnering with Hippocratic AI and Abridge to co-develop products. Other incumbents like Microsoft, IBM, and Amazon joined forces with Mistral AI to make its models available on their cloud platforms. Notably, Microsoft and OpenAI were reportedly collaborating on a USD 100 billion AI supercomputer project to unveil an AI supercomputer named "Stargate" by 2028, likely to be a significant step forward in AI technology.

M&A: Infrastructure disruptors continue to dominate the M&A landscape

Analyst Take: 11 M&A deals were reported compared with four in the previous quarter. Most acquisitions were in the infrastructure space, with Databricks reporting two during the quarter. Companies are gradually expanding their offerings through acquisitions in the GenAI applications space, particularly in marketing content generation. The FMs space is yet to see M&As.

• Data infrastructure and data analytics solutions provider Databricks announced two new acquisitions: 1) Einblick to enhance its GenAI portfolio and 2) Lilac to support data quality efforts for developing GenAI apps. These mark its third and fourth acquisitions within eight months. Also, AI chip startup Groq acquired Definitive Intelligence to scale its cloud platform, while Protect AI acquired Laiyer AI to enhance LLM security.

• Two acquisitions were reported in the GenAI Applications space: marketing content generator Jasper acquired AI image platform Clipdrop from Stability AI to expand marketing capabilities and Typeface acquired TensorTour to expand multimodal capabilities. 

• Notably, Apple also acquired DarwinAI and Brighter AI ahead of a major expansion into GenAI across its products and services reportedly planned for this year.

Appendices

Appendix 1

Funding by hub

Appendix 2

Startups that raised external funding for the first time in Q1 2024

Appendix 3

Disruptor partnerships in Q1 2024

Incumbent partnerships in Q1 2024