Saturday, November 16, 2024

Google manages to keep the cost of AI search flat

Must read

Google has managed to cap the costs it incurs when using AI to generate results.

The search giant created a tool called “AI overviews” that sees search results generated by artificial intelligence presented alongside its conventional list of relevant websites. Such tools are felt to be necessary, given the explosion of interest in generative AI – which Microsoft and OpenAI have promoted as a fine way to find info online.

Generative AI is therefore a colossal challenger to Alphabet’s core search and ads business. That means Google needs to deliver a good AI experience of its own, and keep costs low while doing it – no mean feat, given it has spent decades tuning its compute infrastructure but now must bring dedicated AI-centric hardware into play.

Investors could rightly be worried that doing so would dent margins. But on Alphabet’s Q2 2024 earnings call on Tuesday, CEO Sundar Pichai reassured them that Google has managed to keep the cost of AI overviews flat. He also mentioned ongoing work to reduce latency so that AI overviews arrive promptly, adding “we are focused on matching the right model size to the complexity of the query in order to minimize impact on cost and latency.”

Pichai also told investors that users aged 18 to 24 are showing “even higher engagement” with AI overviews – a welcome outcome given Google is considered an old-school outfit by youngsters reared on TikTok.

The CEO also opined that AI is just getting started. Asked if training new models is becoming harder due to lack of data, Pichai replied that Alphabet is “driving constant progress in terms of the capabilities of the models” and “more importantly, taking them and translating into real use cases across the consumer and enterprise side. I think on that frontier, I think there is still a lot of progress to be had. And so we are pretty focused on that as well,” he explained.

While that work continues, AI is already making a contribution to Google Cloud’s revenues. “Our AI infrastructure and generative AI solutions for cloud customers have already generated billions in revenues and are being used by more than two million developers,” CFO Ruth Porat told investors.

Some of that spend showed up in Q2, during which Google Cloud won $10.35 billion of revenue and delivered $1.2 billion in net income. That’s a 29 percent year-on-year revenue leap, and vast improvement on Q2 2023’s $395 million net income.

Overall revenue was $84.7 billion – up 14 percent. Net income jumped over $5 billion to $23.6 billion, as margins widened from 29 to 32 percent.

Porat made several mentions of “durably reengineering our cost base” – ongoing efforts to trim Alphabet’s workforce, “optimize” its real estate portfolio, and centralize procurement. Keeping the cost of technical infrastructure and ops low also remains on the agenda.

So does buying more of it: the CFO revealed Q2 capital expenditure of $13 billion, which she said was “once again driven overwhelmingly by investment in our technical infrastructure with the largest component for servers followed by datacenters.”

Porat also revealed that Alphabet intends to invest a further $5 billion in self-driving taxi outfit Waymo, so it can “continue to build the world’s leading autonomous driving technology company.”

Despite Alphabet’s earnings exceeding expectations, investors weren’t excited – after spending the day at around $184, its share price spiked to $188.65, but then dropped to under $180 in after-hours trading. ®

Latest article