Top Trends in Generative AI

By
Anthony Sharpe
on
July 19, 2024
Category:
Artificial Intelligence

EDGE100 Report, 2023

Identifying and evaluating early stage startups is challenging for leaders in corporate innovation, corporate strategy, venture capital, and more. Download the report to learn who made it on our list of top 100 startups.
DOWNLOAD

Separating the hype from the hits with a look at what’s really happening in GenAI

Despite headwinds for other sectors, there appears to be no stopping the generative artificial intelligence (GenAI) train. Amid a fall in VC funding across almost all sectors tracked by SPEEDA Edge, work funding increased by $27.7 billion (67%) in 2023, mostly on the back of GenAI-related investments. ChatGPT’s stratospheric success drove $22.7 billion in funding for GenAI startups last year, a 9x YoY increase and almost a quarter of total VC funding raised across all SPEEDA Edge industry hubs during 2023. 

That trend has only continued in 2024, with Amazon’s bold $2.75 billion investment in Claude developer Anthropic, Chinese chatbot developer Moonshot AI’s $1 billion funding found, and infrastructure provider Kyndryl’s $500 million post-IPO debt spearheading a $6.5 billion growth in the GenAI ecosystem in Q1 2024 (up 59% QoQ). 

Of course, we’ve seen funding chase hype before, so it’s important to go beneath the surface to find what’s really happening and what can be expected to stick. 

Key trends in GenAI

  • Rapid growth within the space is seeing a proliferation of partnerships among both startups and incumbents to increase their scope of services and penetrate new market segments. 
  • For similar reasons, M&A activity is heating up too, with more than 10 deals reported in Q1 2024. 
  • Demand for GenAI on smaller or less powerful devices is driving development of small language models, which are faster, more cost-effective, easier to deploy, and require less training data. 
  • GenAI is being integrated directly into hardware to expand its capabilities, offering potential across smartphones, wearables, automotive, cameras, and IoT devices. 
  • Open-source GenAI models have improved greatly, with some predicting they will become comparable to proprietary models in 2024, depending on use case, resources, and available data. 

A flurry of partnerships

The impact of disruptors on the market can clearly be seen in the fact that such companies were involved in more than 70% of the 82 partnerships observed during this period. 

Smartphones are a key driver of these, as incumbents race to incorporate foundation models (FMs) into their devices. Google has found itself on both sides of the equation, working to bring its Gemini AI capabilities to mid-range phones via MediaTek’s Deminsity chips, while also working with Samsung to bring Gemini to Samsung S24 phones via Google Cloud. 

Differing regulations and restrictions require different partners. Apple is looking to leverage Gemini AI on its phones too, but launching GenAI on its phone in the lucrative Chinese market is a different story, which is why the tech giant is in preliminary talks with Baidu around using its tech for China instead. Samsung has gone the same route for its S24 line-up in China. 

On-device partnerships don’t stop at phones, either, with Apple securing Adobe’s text-to-image tool and OpenAI’s ChatGPT for its Vision Pro headset, and Microsoft working with Samsung and OPPO to deliver intelligent connectivity and its GenAI chatbot Copilot to their devices. 

On the infrastructure front, Mistral AI is meeting the growing demand for cloud GenAI capabilities with a host of partnerships, collaborating with major providers like IBM, Amazon, Microsoft, and Snowflake to make its models available in the cloud. NVIDIA, which has been catapulted to being the world’s most valuable company by the AI frenzy, has entered a raft of partnerships aimed at delivering integrated data security, management, and computing infrastructure to incumbents including Amazon, Microsoft, Oracle, and Lenovo. 

The evolving market continues to drive M&A activity.

The innovation of startups and the rapidly evolving GenAI market will continue to drive M&A activity. Databricks, which came to the fore by disrupting the traditional data warehouse market with a focus on analytics, is expanding its capabilities through acquisitions of data science platform Einblick AI and applied research startup Lilac. These follow hot on the heels of its acquisitions of MosaicML and Arcion in 2023, signaling an intention to take its rivalry with legacy software giants to the next level. 

Apple doubled down on its push to incorporate GenAI across its portfolio, bringing its considerable resources to bear by acquiring DarwinAI and Brighter AI.

While M&As have yet to materialize around FMs, the GenAI applications space has seen content platforms broadening their scope through savvy acquisitions, with Jasper AI buying image generator Clipdrop and Typeface enhancing its Multimodal Content Hub by buying TensorTour.

Small language models: push for on-device AI is pushing enormous development

Large language models (LLMs), such as ChatGPT, have received the bulk of the public interest, but the push for on-device AI is pushing enormous development in the field of small language models (SLMs). That’s because the LLMs we’re familiar with use many billions of parameters to output results, which makes them too computation- and energy-heavy for mobile devices. 

SLMs offer a more streamlined version of their larger cousins, operating on fewer parameters, and requiring less training data and time to become viable. This makes them pretty desirable for mobile applications, but also for smaller businesses without the budgets or time to develop larger models, and those with specific applications for which a fine-tuned model is more appropriate. That smaller footprint also means less code and thus fewer security vulnerabilities – crucial considering the breakneck pace of enterprise GenAI development and deployment. 

Examples include Microsoft’s Phi series, which operates on “only” a couple of billion parameters, Google’s open-source Gemma models, and Mistral’s 7B. Meta is going big on small models, having recently unveiled its MobileLLM and continuing research aimed at reducing the amount of AI-related data processing that takes place in data centers. This has major implications for the energy and cost efficiency of GenAI down the line. 

AI in your Ears: Hardware integration

These smaller models are enabling greater integration into a range of devices beyond smartphones. The implications for wearables, hearables, and other small devices are immense, with companies only scratching the surface of what is possible. 

Google is reported to be looking at incorporating Gemini into its earbuds, while the aforementioned ChatGPT and Apple Vision Pro are also noteworthy examples. IoT devices, which continue to proliferate in every aspect of our lives, also stand to be enhanced, with the potential not just to gather but also process and analyze increasing volumes of data. 

The power of community collaboration: Open-source models

There’s no denying the fact that OpenAI brought GenAI kicking and screaming into the public consciousness, kickstarting a development and funding arms race. However, open-source software enables a genuine proliferation of re-examination, redesign, and evolution through the power of community collaboration. Until recently, open-source GenAI models had lagged behind their proprietary counterparts, but recent developments have leveled the playing field considerably. 

These developments have come from incumbents like Mistral AI and Stability AI, which have launched powerful open-source models, as well as from Meta, which in February announced that it was merging two of its AI research teams to work on artificial general intelligence, with a commitment to keeping this development open-source. 

Stay on top of these trends

It’s clear that GenAI will continue to develop at breakneck speed, making well-informed market intelligence increasingly essential for businesses across every industry. SPEEDA Edge aligns corporate strategies with their partnerships, acquisitions, and investments in emerging industries. By visualizing and comparing these activities, we help guide strategic priorities.


SPEEDA Edge not only gathers a vast amount of information on industries, technologies, and companies, but also ensures that all data on our platform is verified, making it the most reliable source for companies needing to understand their competitors, identify current and future use cases, and make well-informed decisions in a highly competitive environment.


Moreover, SPEEDA Edge's team of analysts is available to assist enterprises in organizing and extracting insights from data. This underlines the fact that, even in the age of artificial intelligence, human expertise and connection remain invaluable.

Conclusion

GenAI is evolving in ways that are sometimes surprising. Organizations across the spectrum are discovering its potential, along with the fact that no two use cases are identical. SPEEDA Edge helps companies understand technological impacts like this and how they affect the competitive landscape. To understand the potential impact of our services on your business, contact us for a personalized demo. 

For more information about the tech trends shaping strategic priorities, head over to our blog and subscribe to our newsletter.

Tags:
No items found.
Anthony Sharpe

Freelance journalist, editor, proofreader, and travel photographer with more than 13 years of experience creating, curating, and improving content.