The rollercoaster of enterprise AI adoption

January 22, 2025

Good morning. Shares of Netflix spiked last night after the streaming giant reported earnings that made Wall Street happy.

But here’s the crazy figure: Netflix has surpassed 300 million paid subscribers. That’s a little less than the population of the U.S.

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

🧠 AI for Good: Artificial hearing
⚕️ DeepMind says AI drugs are going to trial this year
🏛️ Trump repeals Biden’s AI executive order
📊 The rollercoaster of enterprise AI adoption

Source: Created with AI by The Deep View

There’s a lot about the brain that we don’t know. But, when it comes to the ways in which the human brain processes auditory signals, here’s what we do know: when a sound wave reaches the inner ear, neurons pick up on the vibration and alert the brain by emitting spikes (rapid changes in voltage).

Auditory neurons, according to MIT, are capable of firing hundreds of these spikes per second, enabling us to recognize voices and react to familiar sounds, among other things. In addition to firing spikes, these neurons are able to time their spikes “with exquisite precision to match the oscillations of incoming sound waves.”

What happened: Leveraging machine learning technologies, researchers at MIT’s McGovern Institute for Brain Research were able to determine that the timing of these neural spikes is actually a vital component of auditory processing.

While researchers have suspected this for a while, it’s been difficult to study. So a team led by MIT professor Josh McDermott developed an artificial neural network designed to simulate the parts of the brain that receive input from the ear.
After validating that the model performed similarly to the human ear, the researchers adjusted the timing of those neural spikes in the fake ear, finding that the ability to identify voices “is lost without precisely timed signals.”

The implications: The research being undertaken here, according to McDermott, will enable scientists to better understand auditory behavior and the details of person-to-person hearing loss, potentially enabling better diagnoses and, in turn, more advanced, fine-tuned hearing aids.

“If you want to design a prosthesis that provides electrical signals to the brain to reproduce the function of the ear, it’s arguably pretty important to know what kinds of information in the normal ear actually matter,” he said.

Elf Labs has opened one final investment window at their current share price.

Why? Major enterprise partnerships are on the horizon that could significantly impact their valuation. We're talking about a company that:

Secured 100+ historic trademark victories for some of the highest-grossing in history like Snow White, Cinderella, Little Mermaid and more
Is powered by next-gen technology protected by 12 patents for AR/VR tech (no headsets needed)
Is launching three major franchises in 2025, backed a licensing team that has done over $6B in transactions

This extension is your final opportunity to invest at the current share price before these enterprise deals are announced.

Secure your 15% bonus shares in Elf Labs here.

Source: Isomorphic Labs

Demis Hassabis, the CEO of Google’s DeepMind — as well as the DeepMind spinoff, Isomorphic Labs — said Tuesday that “we’ll hopefully have some AI-designed drugs in clinical trials by the end of the year.”

The details: Hassabis, who shares a Nobel Prize in chemistry for DeepMind’s work on AlphaFold, a protein prediction AI model, was speaking on a panel at the World Economic Forum in Davos. “That’s the plan,” he said.

Isomorphic Labs spun out of DeepMind in 2021, just a few years after DeepMind launched the first version of AlphaFold. It functions as an autonomous subsidiary of Google parent Alphabet intended exclusively to build on AlphaFold’s breakthroughs in protein folding.
The company is, in short, on a mission to use AI to accelerate drug discovery.

Last year, Isomorphic Labs secured partnerships with two pharmaceutical companies — Eli Lilly and Novartis — to collaborate on small molecule research. These partnerships marked the company’s first pharmaceutical team-ups.

It’s not clear what specifically the lab is working on that is on the brink of entering clinical trials — Isomorphic Labs’ website includes no pipeline information, and the partnerships with Eli Lilly and Novartis are focused around undisclosed targets. But Isomorphic Labs said these partnerships have the potential to be worth a minimum of $3 billion (excluding royalties from drug sales).

Isomorphic Labs is far from the only AI-driven biotech firm focused on drug discovery — it’s an approach also being utilized by Lantern Pharma and Recursion Pharmaceuticals, both of which have several drugs creeping up the stages in their pipelines.

Supercharge your organization's AI adoption journey from complex challenges to strategic success with Tines' must-read guide: "Securing AI in the Enterprise." *

The global struggle over how to regulate AI (Rest of World).
Global climate finance alliances at risk as top lenders pull out (Semafor).
Live updates on Trump's executive orders and actions on first full day of second term (CBS News).
Life is thriving in the subsurface depths of Earth (Ars Technica).
Cement has an emissions problem. Can tech that mimics coral fix it? (Canary Media).

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

The U.K. has made a new package of AI-powered tools — nicknamed ‘Humphrey’ available to government employees in an effort to speed up the rate of public service.
CBS reported that President Trump is expected to announce a multi-billion-dollar private sector investment to build AI infrastructure in the U.S. OpenAI, Softbank and Oracle are all reportedly involved in the venture, coalescing around a project nicknamed ‘Stargate.’ This announcement has come out; we’ll be sinking into it tomorrow.

Source: White House

Mere hours after taking office, President Donald Trump — exactly as expected — revoked former President Joe Biden’s circa-2023 executive order on AI, ‘The Safe, Secure and Trustworthy Development and Use of Artificial Intelligence,’ (in addition to a long list of around 70 other standing executive orders).

The details: It’s not yet clear if Trump plans to replace the order with one of his own, or what the contents of that executive order might hold. Last year, the official Republican platform claimed that the order hindered innovation, and said it would be replaced with one “rooted in free speech and human flourishing.” Further details have not been forthcoming.

Though lacking in clear methods of enforcement, Biden’s order called for companies to share the results of safety tests with the U.S. government before a public release. The order also sought to establish clear consumer and civil rights and protections related to AI, while also pushing for government adoption.
The order — alongside a national security memorandum more recently published by Biden — marked the only real federal levers for some semblance of AI regulation in the U.S.

The revocation comes despite Elon Musk’s position in the Trump administration. Musk has previously discussed the safety risks posed by AI, going so far as to endorse California’s controversial SB 1047 last year. Lately, though, he’s been pretty quiet on the AI safety front.

Marc Andreessen, however, who runs the tech VC firm Andreessen Horowitz (a16z), is reportedly recruiting and placing staff across the Trump administration. Andreessen is very publicly against the idea of AI regulation. And former a16z general partner Sriram Krishnan will serve as Turmp’s senior AI policy advisor.

Source: Unsplash

The biggest ongoing bet in the business of artificial intelligence doesn’t really have much to do with science fiction, HAL 9000 or superintelligence. It’s much more simple. The bet is enterprise adoption of current generative AI tech.

For the developers and labs spending hundreds of millions of dollars on developing and deploying their tech, the enterprise promises a gold mine of value derivation. That’s why we’ve seen such a universal focus on pushing enterprise AI adoption ever since this AI craze began in 2023.

But adoption has been uneven. The rash excitement that flooded executive suites and earnings calls in 2023 was blunted somewhat in 2024, over unsolved concerns related to reliability, usability and cybersecurity (as well as cost). Despite this, adoption grew significantly in 2024, with one report finding that corporate spending on AI grew from $2.3 billion in 2023 to $13.8 billion in 2024.

But the rollercoaster ride of enterprise adoption is far from over.

What happened: Deloitte on Tuesday published its fourth quarterly State of Generative AI in the Enterprise report, a survey of nearly 3,000 C-suite-level executives across 14 countries. The report identified a broad mindset of “positive pragmatism” going into 2025.

Key findings: The report found that adoption is still moving along slowly at individual corporations. The tech remains limited to less than 40% of the workforce; less than 60% of those who have access to it use it on a daily basis. A majority of respondents, meanwhile, are still firmly in the experimentation phase, saying that they expect 30% or fewer of their AI experiments to fully scale into operation within the next six months.

Despite this slow pace, 78% of those polled said they plan to increase their spending on AI initiatives over the next fiscal year, a theme that has been ongoing for several years, now.
Today, according to the survey, the most advanced and impactful generative AI applications involve IT, with only 10% of respondents using genAI for marketing and 8% for cybersecurity.

The ROI: The return on the investment of these AI initiatives, however, remains a major question mark in the industry. And according to the report, actual, measurable ROI is actually beginning to hit the books.

74% of respondents said that their single most advanced generative AI initiative is meeting or exceeding their ROI expectations; 41% reported ROI related to their most advanced initiative between 11% and 30%.

Those applying genAI to cybersecurity are currently seeing the biggest ROI, according to the report.

This obviously doesn’t examine the cost and ROI related to a company’s full slate of AI experiments.

It’s not all rosy: In the midst of the very beginning of a corporate transition to agentic AI, the report highlighted a number of concerns that corporations expect will further slow their pace of adoption.

35% said that mistakes leading to real-world consequences will slow things down; 30% highlighted a lack of quality data; 29% pointed out a possible lack of trust due to hallucinations and biases and 25% called out copyright concerns.
38% said that concerns over regulatory compliance represent the single biggest barrier to enterprise integration of generative AI.

“Amid the promise of AI agents and the evolution of foundational models, future-thinking organizations are as bullish as ever in building bridges to ROI, all while understanding the need for nuance — and patience — as we embrace this next wave of GenAI,” Jim Rowan, Deloitte’s Applied AI leader, said in a statement. “Anticipation is high, and now is the time for leaders to take the long view of their GenAI investments, with a focus on governance, collaboration and continued iteration as key accelerators in the race for sustainable value.”

This echoes themes that have been developing recently, that AI is a long-term play, not a short-term bump.

This push for function-specific tasks does represent a departure from the early days of adoption, during which the tech was perhaps best described as a solution in search of a problem. Companies are seemingly taking a more practical approach now, an approach that might present an unwanted challenge to developers.

Two things are occurring right now at opposite ends of the same spectrum. One, the enterprises are narrowing their approach, and focusing in on direct paths to ROI. And two, the developers — burning millions or billions of dollars — are looking for ways to start making some of those VC dollars back (some by offering more expensive subscriptions).

So we have an industry looking to charge more, at a time when businesses are looking to either pay less, or determine legitimate economic value relevant to those higher price tags.

These two things seem to me to be in conflict.