Editor’s note: In the second of a two-part report, Technology Business research Analyst Jennifer Hamel examines what the ascent of artificial intelligence means to information technology providers like IBM – software, hardware and services.

HAMPTON, N.H. – The implications for vendors from the ascent of artificial intelligence across the IT landscape: From software to hardware to vendors:

  • Software

In the rush to make a market around AI, IBM came out of the gate in 2011 with bold assertions around the transformative impact Watson would have on businesses and on society as a whole. Over the last six years Watson has evolved from the monolithic computer that beat the top two winners on the game show “Jeopardy” to a brand name for a diverse stack of data services, cognitive APIs, and industry- and domain-specific applications. This shift improved IBM’s ability to monetize its IP by clarifying Watson’s value for customers in solving specific business and technical challenges. Newer AI software vendors watch Watson closely — even as they insist their technologies are more advanced or easier to use — and clearly benefit from the market noise IBM has created to commercialize its cognitive innovations.

Now, AI startups and established software players building AI capabilities similarly structure their portfolios and go-to-market messaging to connect potential buyers to AI solutions at the appropriate level of abstraction for their business need and technology maturity, providing accelerators and design thinking services as needed to put buyers on the path to a successful AI solution. As this type of persona-based portfolio development becomes table stakes, we expect the next battleground for AI software vendors will be around model interpretability — that is, enabling non-expert human users of AI systems to understand and be able to explain machine-generated results.

This will be critical for AI to become truly integrated into business decisions, particularly in highly regulated industries, as companies will need to break into the black box of the algorithm to build confidence in their AI’s recommendations.

  • Hardware

AI’s intense data processing requirements challenge chip vendors to innovate, acquire and partner strategically to stay competitive. According to TBR’s 1Q17 Intel report, Intel (NasdaqGS: INTC) aims to build standards and baseline technology for AI, which it hopes to disseminate across the technology landscape. TBR views the recent establishment of the Artificial Intelligence Products Group, headed by former Nervana Systems (which Intel acquired in August 2016) CEO Naveen Rao, as one step of many Intel has taken to become an AI market-maker.

Rao’s keynote at the conference highlighted another key step, the beta launch of the Intel Nervana Graph, which enables developers to work with deep-learning frameworks across various hardware platforms, including Intel’s CPUs and competitor NVIDIA’s GPUs.

NVIDIA also appears to be building an ecosystem through partnerships with software vendors to create GPU-powered AI solutions tailored to specific customer needs, such as driverless cars (H20.ai), digital marketing (SAP [NYSE: SAP]) and real-time big data visualization (MapD). Startup Mythic looks to differentiate by redesigning the chip itself to eliminate the processor, enabling AI computation to be run inmemory within individual devices to avoid the latency and security issues of a cloud environment.

TBR will continue to monitor how vendors adapt to varying hardware needs for AI (e.g., silicon versus quantum chips, cloud versus edge computing), but we expect vendors that offer flexibility in deployment methods will have the most longevity.

  • Services

Similar to the sentiment observed at the Collision 2017 event, startups at the O’Reilly AI Conference did not appear to rely on traditional consulting firms or systems integrators to provide channels to market, despite the efforts of many of those firms to build startup ecosystems as part of digital transformation initiatives. Several of the AI startups TBR spoke with work directly with their customers through dedicated support and professional services teams and have yet to create robust partner programs.

One reason that was fairly common across startups was the desire to ensure customer success using the startups’ tools or platforms, of which a third party would be unlikely to have as intimate of knowledge as the startup. Other reasons that came up were perceptions that traditional services firms are too difficult to work with and lack technical execution skills around AI.

We expect as startups evolve from proving their technology works to building scale, partnerships with services vendors such as Accenture (NYSE: ACN), Deloitte and Cognizant will be more attractive to the startups, particularly as these vendors bring industry, change management and IT expertise that AI technology startups lack. However, we expect boutique consultancies that emerge with AI at their core and hire industry talent to provide strategy and process advice in discrete verticals will pose a threat to these firms.

For example, London-based ASI Data Science offers AI and machine learning solutions for financial services, retail and government and helps organizations build internal data science teams through recruiting and training programs, rather than angling for long-term managed services opportunity.

TBR perspective

Though there were few answers to the big questions of AI’s societal impact at the O’Reilly AI Conference, the event moved the conversation in the right direction — away from AI as a mystical black box or a means to robotic world domination, and toward AI as a tool for research breakthrough and business transformation.

Vendors across the IT spectrum must continue to evolve their strategies and messaging to accelerate customer adoption, including:

 Understanding customers’ needs and business challenges before introducing AI as a solution

 Setting realistic expectations for what AI can and cannot do today

 Dealing with organizational and regulatory impacts head-on

 Building partnerships and ecosystems that enable customers to experiment with various AI technologies

 Working with research organizations and policymakers to try to resolve broader issues such as bias, education and the future of work.