Industry executives and experts share their predictions for 2025. Read them in this 17th annual VMblog.com series exclusive.
How the move to a
platform-approach requires the rethinking of tooling to support two distinct
sets of priorities.
By Mark Fussell – Co-founder and CEO of Diagrid and co-creator of the Dapr open-source project
The discipline of distributed applications –
whether you call them microservices, cloud-native architectures, or something
else – is seeing the rapid adoption of a platform
approach: the establishment of dedicated platform which engineering teams
that are, in turn, tasked with ensuring security, scalability, performance, and
cost efficiency across all development efforts in the organization. The end
goal is to increase developer velocity and productivity by allowing the
platform team and the application development team to specialize in separate
concerns, achieving a balance between repeatable, shared best practices, and
the adoption and rapid deployment of new technologies.
Achieving all this simultaneously? Easier said
than done.
Most organizations are grappling with a
patchwork of legacy codebases while racing toward the cloud. However, despite
all of its attractions, moving applications to cloud platforms – especially
public cloud services – often brings an unwelcome surprise: runaway costs.
Simultaneously, growing concerns about data
sovereignty and confidentiality are pushing organizations to rethink their
strategies. These dual pressures are driving a noticeable trend toward hybrid
infrastructure models, combining cloud with private data centers to regain
control.
Platform
modernization is the order of the day, but nearly
everyone we talk to feels under-resourced. The need to move quickly while
adopting cutting-edge technology has never been greater. As we look ahead to
2025, here are five key predictions that we see shaping the future of
cloud-native platforms.
Prediction #1: Platform Teams Will
Emphasize Consistency and Cost Efficiency
The rise of platform engineering is
transforming how organizations build and manage their development environments.
By establishing consistent, repeatable processes, platforms can enhance
developer efficiency while keeping costs in check. As we move into 2025, we
anticipate broader adoption of a platform approach intended to streamline
operations, reduce variability, and ensure predictable outcomes for both
development and deployment.
How well the platform team interfaces with the
app developers will be a major factor in whether or not this approach succeeds.
Prediction #2: The Era of FinOps and
Enhanced Visibility
With cost control becoming a critical
priority, organizations are investing heavily in infrastructure visibility and
observability. This includes real-time monitoring of resource utilization,
spend, and performance metrics. FinOps
– a discipline that brings financial accountability to cloud spending – will
gain momentum. Platform teams will increasingly integrate cost observability
into their tooling to prevent overruns and ensure that infrastructure remains
both performant and cost-effective. This approach will extend to permissions at
the team level, meaning that developers will have restricted access to
expensive infrastructure components.
We expect teams to invest more time in
performance optimization to get more value from their current spend.
Prediction #3: Security and Data
Sovereignty Remain Center Stage
As attacks grow more sophisticated and
regulatory frameworks evolve worldwide, cybersecurity and data governance will
remain primary concerns. Whether workloads run on public clouds or private data
centers, organizations are demanding tighter control over their infrastructure.
This includes managing network layers, firewalls, and enforcing strict data
sovereignty policies. Platforms that can offer comprehensive security and
compliance features at both the user and the service level will become
indispensable.
New regulations, such as the European Union’s Cyber Resilience Act (CRA) have now come into effect, and teams have a
hard deadline to meet in order to become compliant.
Our prediction here is that application
development and platform teams will be dedicating more time in 2025 to the
auditability and compliance of their systems.
Prediction #4: AI Integration into applications
Without Breaking the Bank
The rise of AI introduces the ability to solve
novel problems more easily such as image recognition, automated agentic
systems, however raises a unique set of
challenges. AI workloads are notoriously compute and data intensive, with the
potential to skyrocket cloud or private data center costs. In 2025,
organizations must navigate how to seamlessly integrate AI into their solutions
without sacrificing cost efficiency. The challenge lies in balancing AI-driven
innovation with the realities of infrastructure limitations and budgets.
Our prediction for 2025 is that there will be
an increase in the number of smaller, more specific AI models that have trained
with company’s data and for targeted scenarios, over larger more general
purpose models. While tooling will
rapidly evolve, application platform teams will want to standardize the use of
AI models based on a careful cost/value analysis.
Prediction
#5 The rise and establishment of agentic AI applications
This is an application development trend that will gain more of a foothold with
application teams. There has been plenty of focus on the rise of agentic
systems and we see this as little different from solving the hard problems of
distributed application development, both with workflows where LLMs and AI
tools are orchestrated and agents where LLMs dynamically direct their own
processes and tool usage determining how to achieve and complete tasks.
Our prediction is that there will be a rise in
building agentic systems to automate common tasks that will range from using
simple single LLM API calls stitched together, to the use of workflow engines
to orchestrate the more complex multiple LLM API interactions. There will also
be more agentic frameworks appearing, but many of these are reinventing already
established distributed computing problems and may add unnecessary complexity
to your applications, so choose carefully.
Dapr: an
insight into distributed application trends
At Diagrid, we see these trends reflected in
our work supporting the open-source Dapr
project, which achieved graduated status within the CNCF at the end of 2024.
With contributions from over 3,900 developers across 400 organizations and more
than 70 million SDK downloads, Dapr has proven its value to the cloud-native
community.
The rapid adoption of Dapr is driven in part
by its ability to bridge the gap between application developers and platform
engineers by addressing their unique concerns while fostering collaboration.
Dapr has also been battle-tested in thousands of production systems solving
many hard distributed system challenges that save weeks and months of
development time.
For application
developers, Dapr accelerates the development process by abstracting away
complexities like service-to-service communication, state management, and
pub/sub messaging. It enables developers to focus on delivering features and
functionality without worrying about reinventing the wheel or accumulating
technical debt. The incremental adoption model is particularly beneficial,
allowing developers to integrate Dapr’s capabilities gradually into their
existing solutions without requiring a full system overhaul. Dapr, with its
workflow and conversation APIs, is positioned to build agentic applications.
For platform
engineers, Dapr provides a consistent and standardized contract to
application management, ensuring that best practices are enforced across the
board whilst providing infrastructure flexibility and multi-cloud support. By
implementing robust security mechanisms, Dapr establishes clear boundaries for
inter-application communication and access control, reducing the risk of
misconfigurations or vulnerabilities. This not only simplifies governance but
also enhances the overall reliability and scalability of the platform, giving
engineers the tools they need to maintain efficient and secure infrastructure
while supporting developer productivity.
With the additional capabilities of Diagrid Conductor, application
platform teams can monitor performance and implement best practices to greatly
reduce the cost of running their applications in production.
As platforms evolve to meet the challenges of
2025, Dapr stands ready to help organizations achieve velocity, maintain
security, and optimize costs-all while empowering developers to focus on
building innovative solutions including those that are exploring the world of
agentic systems for improved automation.
Here’s to a transformative and exciting year
ahead.
##
ABOUT THE AUTHOR
Mark Fussell is the CEO and co-founder of
Diagrid. With over 30 years of experience, Fussell has a track record of
building innovative computing platforms, managing large-scale cloud services,
cultivating open source communities, and implementing successful go-to-market
business strategies. Previously at Microsoft for more than 20 years, he helped
develop the distributed systems platform for Azure’s hyperscale services. In
2019, Mark co-created the Distributed Applications Runtime project, known as
Dapr, one of the fastest-growing open source projects centered around
application development and a graduated project of the Cloud Native Computing
Foundation.
