Google Cloud Startups leader sees opportunity for startups building AI agents amid pressure to move quickly

Darren Mowry, managing director at Google Cloud Startups, is excited by the opportunity that exists in artificial intelligence (AI) right now but said this also comes with a lot of responsibility.

Mowry leads Google Cloud’s startup-focused growth and acceleration strategies globally. In Canada, Google supports startups through Google Cloud’s services and the company’s accelerator programs, including the Google for Startups Accelerator, which has catered to more than 100 early-stage Canadian tech companies over the past five years.

Google Cloud’s Canadian tech customers include Toronto-based large language model (LLM) developer and Open AI competitor Cohere and Montréal-based online mortgage lender Nesto.

“There’s going to be thousands of agent providers and AI-enabled applications. And that’s really where startups are finding their legs.”
Darren Mowry,
Google Cloud Startups

In an interview with BetaKit ahead of the company’s latest Google Cloud Next Conference, Mowry noted that 18 months ago, developing AI chips and models was all the rage. Today, he said that startups’ focus has largely shifted towards building AI applications and agents.

Mowry argued that there is so much opportunity in AI right now that “doing the right thing in the right order” has become a key challenge for Google Cloud and other companies, one that has forced the cloud services giant to be selective and prioritize where it invests.

“We could go do 1,000 things … but what we’re going to do is pick the smaller number of things that we believe are going to have the maximum value for enterprise customers, for public sector agencies, and for startups,” Mowry said.

But the global AI race and tech’s desire to move quickly has also led to missteps. Given the power and scale of the nascent tech, whether firms are responsibly applying it with the right guardrails to protect against bad outcomes is one of the things that keeps Mowry up at night.

“AI is a technology that if misused, the results can be something … that no one is looking forward to,” Mowry said.

The rollout of AI has not always gone perfectly for Google. Take, for instance, Google’s AI search results recommending that users put glue on pizza.

RELATED: Google has a glue problem

“In the early days, you may remember moments that—believe me, my head was in my hands—as there were articles written about Bard saying something or early Gemini saying something else,” Mowry said.

These problems have not been fully solved. Yesterday, Google’s AI search results informed The Globe and Mail’s technology reporter Sean Silcoff that he was the CEO of Toronto corporate virtual private network unicorn Tailscale.

Google’s rush to win the AI race has led to ethical lapses, including making compromises on misinformation and other harms in order to catch up with OpenAI’s ChatGPT, according to employees who spoke with Bloomberg, and expressed concern concern that the speed of AI development is not allowing enough time to study potential risks.

Meanwhile, the American search giant’s parent company, Alphabet, also recently lifted a ban on using its AI for weapons and surveillance—an application that even Edmonton-based AI doomer denouncer, former Alphabet DeepMind leader, and recent Turing Award winner Richard Sutton has expressed wariness about.

For his part, Mowry said he is ultimately glad that he is working at Google given the firm’s approach to responsible AI. Mowry claimed that Google intends to continue investing “hard and fast” in AI, while also working to ensure that it is deploying the tech responsibly, and questioned whether “everyone is being as responsible as we should be” when it comes to AI.

Amid the AI boom, Google Cloud has been trying to position itself as the place to build for AI companies. Today, Google Cloud claims that 90 percent of generative AI unicorns and over 60 percent of funded generative AI startups are clients.

At Cloud Next, Google made a slew of AI announcements, introducing what it claimed is the most powerful chip it has ever built—an AI tensor processing unit for running AI models and applications—as well as a new low latency and cost-efficient thinking model, some fresh automation features across its office software products, and ways for organizations to create and manage AI agents aimed at helping its clients use the tech in a variety of ways.

Many AI startups have been generating eye-watering growth, but some investors have expressed doubts that all of these companies will be able to maintain their surging revenues amid fears about differentiation.

Mowry believes that we are not close to hitting an AI plateau. “People are almost talking about AI as if we’ve hit a plateau,” he said. “I can promise you … we’re nowhere near the plateau of AI yet. I think we’ve got a lot more rocketing to do.”

“The amount of climate tech innovation that’s coming from Canada puts the US and other markets to shame.”

Mowry noted that he is seeing sustainable business value being created in areas like biotech and climate tech.

“The amount of climate tech innovation that’s coming from Canada puts the [United States] and other markets to shame,” Mowry added.

Mowry also cited the use of AI to make technologists more productive and help with content creation—something Mississauga-based social media agency and Google Cloud customer Viral Nation is working on—as two other applications that he believes possess staying power.

As to the broader shift that has taken place from startups focusing their efforts on building AI chips and models to developing applications and agents, Mowry noted that many companies have realized they are better served by solving a specific, differentiated problem with AI and letting others pour billions of dollars into building “the core plumbing.”

“At the end of the day, there may be a handful of chip and infrastructure providers, there may be dozens of model providers, but there’s going to be thousands of agent providers and AI-enabled applications,” Mowry said. “And that’s really where startups are finding their legs. That’s where the venture capital money is going.”

Feature image courtesy Google Cloud.