AI Highlights — the whole picture — 2026-10-21 (Wed)

The momentum of AI investment can no longer be measured solely by the amount of equipment. While some believe that the US economy is close to recession, excluding AI investment, HSBC's CEO warned that there is a need to be selective about data center investments. During the same period, Capable raised $38 million for its drug discovery lab and Disney spent $180 million on talent for Character.AI. Funds have expanded from expanding computing resources to partnering with others, human resources, and uses. What this half-day article has in common is that rather than looking at the size of the investment amount, it is necessary to look at what the funds are tied to and under whose judgment and responsibility. In this text, we will examine each aspect of investment, partnerships, human resources, regulations, and creation in order.

Selective eyes will be placed on investments that support the economy

Some people are questioning the way they choose investments based on their view of the U.S. economy. CFO.com reported, ``The U.S. economy is close to recession, excluding AI investment.'' This view shows that the support for the economy is biased toward AI investment. HSBC's CEO warned on CNBC that even amid the AI boom, data center investments need to be selective. The same atmosphere exists in the market. Seeking Alpha reports that concerns about AI have reignited, interest rates have expanded in peripheral countries in Europe, and oil prices have continued to rise. While AI investment supports the economy, momentum can only be measured by the size of the investment. However, since there have been simultaneous calls for doubts about its support and calls for selection, readers need to look at what the investment will lead to, rather than the size of the amount.

Funds are going beyond equipment. Capable has raised $38 million in seed funding to build an AI drug discovery lab, according to Dealroom. Apart from equipment expansion, we can see that new financing is being raised for specific purposes. On the major side, Yahoo Finance reports that AstraZeneca (LSE:AZN) could be undervalued by 22% based on AI partnerships and pipeline updates. This number is an estimate based on the analysis of the same article. Here, companies are evaluated based on how they use AI and which alliances and development pipelines they link it to.

What all four ingredients have in common is that the categorization of AI alone is becoming insufficient as a reason to collect funds or raise evaluations. The units that need to be looked at are detailed, such as the scale of the data center, the purpose of drug discovery, and the partner companies of major companies. Once we enter the stage where the nature of the investments that support the economy is being questioned, the next thing to look at is where the funds are being transferred from equipment.

Pharmaceutical alliances are reselecting scale and partners

Eli Lilly has partnered with METiS Therapeutics on AI-powered delivery technology. According to BigGM Finance, the down payment is over $10 million, and milestones are up to about 170 billion yen. FourWeekMBA reports the deal as "an AI contract worth up to $1.1 billion." 36Kr is positioned as the first overseas export project for a large-scale generation AI-LNP platform originating from China. The same article also mentions a change in Lilly's idea of "bulk acquisition".

Around the same time, AstraZeneca began a global AI collaboration with Oracle to optimize clinical trials. AstraZeneca's announcement, reported by Reuters. FourWeekMBA juxtaposed this announcement with Lilly's contract in one article, ``AstraZeneca chooses Oracle, Lilly signs AI contract worth up to $1.1 billion.''

Public funding is also included in drug discovery AI. According to citybiz, Sinopia Biosciences has received a $2.5 million grant from the National Institutes of Health (NIH) to advance AI drug discovery. The contract between companies is worth up to $1.1 billion, while the public grant is $2.5 million. Even if the amount is small, it shows that the money is going to the same area.

The three moves have one thing in common in how they choose their opponent. Lilly chose an outside company with strong delivery technology. AstraZeneca will partner with Oracle to develop the infrastructure used to operate clinical trials. The system of developing infrastructure solely in-house is expanding to the system of combining it with external partners for each application. The total value of the contract is a maximum value subject to the achievement of milestones. The headline number should be read as an upper limit, as the only confirmed payment is the down payment of just over $10 million.

What readers should look at is who was assigned to which process, rather than the size of the amount. The selection for each process also depends on who is providing the funds. In the next section, we will look at how this selection extends outside the company, to human resources and public support.

Redistribution of human resources and roles will take precedence

According to Cartoon Brew, Disney spent $180 million to retain talent from the bankrupt Character.AI. Army Times reports that the U.S. Army Cyber Command is searching for a balance between AI and humans. The former is a move by companies to acquire human resources, and the latter is a move by organizations to re-determine the division of labor between people and AI. Although the formats are different, in both cases the focus of funding and discussion is not on the scale of the equipment, but on who is responsible for what work. Readers looking at AI investments should check not only the amount of increased computing resources, but also the amount of human resources acquired and whether there is a policy for determining the division of labor between people and AI. The $180 million figure shows that human resources themselves are valued as investments.

At the same time, moves have been made on the administrative and educational sides. According to Patch, the state of California has released a job search support tool called ``Skill Matching'' that uses AI. The state has created a system that connects people looking for work with the skills they need. According to the Herald Democrat, Grayson University is considering AI literacy and engineer training courses. State retraining support and university training programs are aimed at both the time when workers are replaced and the time when new leaders are being trained. Not only corporate recruitment and military operations, but also public systems are beginning to change to match the allocation of people.

Tom's Guide asks whether giving chatbots and AI agents a face increases trust. If the division of labor between humans and AI is to be rearranged, the extent to which users view AI as a partner they can entrust will determine the division of labor. In the examples so far, different actors are working on each layer: Disney and the Army Cyber Force for deployment, California for retraining, Grayson University for education, and Tom's Guide for contact design. What readers should look at is whether these are aligned in the same direction, rather than the size of a single announcement.

If the location of people is decided first, then the next question is what uses the actual funds will go to based on that location.

National delineation and safety design

The United States has banned DeepSeek, citing "threat vectors" as the reason, Fast Company reported. China has also responded. Around the same time, The Decoder reported that Altman presented three proposals for AI security, including one that would shut down other companies' GPUs. The former is about the government drawing a line that prohibits the use of certain AIs, and the latter is about whether businesses have the authority to stop other businesses' computing resources. Although the actors are different, the question is the same: who can run the AI and who can stop it. In addition to the performance and price of the technology, readers who choose a source for procurement or partnership will need to consider not only the performance and price of the technology, but also which country's regulations and which company's suspension authority the supplier is subject to.

The line is tested in the field through trust and fairness. According to Diagnostic Imaging, the FDA has cleared AI software that analyzes MRIs to assess the prognosis of mild cognitive impairment. This is a case where regulatory authorities have specified and approved the use, and users can consider introducing it based on the scope of the approval. Meanwhile, WDET 101.9 FM covered the online conference discussion surrounding AI in healthcare and racial bias. Being accredited is one thing, working fairly for all patients is another. Rather than deciding whether to adopt or reject a project based solely on whether it has approval or not, it is necessary to confirm whose data was used to verify the data and which groups might have differences in results.

When we line up the four ingredients, the points of prohibition, suspension authority, authorization, and prejudice all boil down to the same design problem: where to place the ``actor'' and the ``stopper.'' Both Altman's proposal and the bans in each country are currently at the proposal and reporting stage, and it is unclear to what extent they will be put into practice. The investments, partnerships, and selection of human resources that we saw in the previous section take place within this line. Choosing where to invest money and people also means choosing which rules and safety mechanisms that destination will follow. In the next section, we will look at how this selection is taking shape on the business side.

Authenticity tested through creativity and relationships

Chicago's queer-led film festival Dare To Dream refuses to submit "art" made with AI (Block Club Chicago). Around the same time, Jenna Ortega said, ``I don't need an AI friend,'' and a movie depicting an AI friend has also become a hot topic (Martin Cid Magazine). The film festival drew the line at the entrance to the film, with actresses stating that they would not have AI friends in their lives. The movie depicts the AI friend as the subject of the story. The answer to whether or not to accept AI as a creator or partner depends on the situation. When deciding whether to entrust creation or social interactions to AI, readers will first check what the environment treats as real rather than the performance of the technology.

On the relationship side, there are clear moves to anticipate demand. The Universidad de Córdoba (UCO) website has an article that lists how to create an AI boyfriend or AI girlfriend, Japanese support, and recommendations for men. From the title, it can be seen that AI Lover is a product aimed at Japanese users. On the other hand, believers ask the question from a different angle. There is an essay on how believers should approach the "welfare of AI" theory (Modern Reformation). What is at stake here is the boundary between the responsibility of those who treat AI with care and whether it is okay to place AI in the same position as humans. From the user's perspective, the choice to use AI as a friend or lover involves a judgment about how to view AI as a being.

Attempts to decipher plant signals are also an extension of the same question. An AI startup has reportedly deciphered the "language" of plants from electrical signals (The Tech Buzz). However, whether or not signals can be called words depends on the interpretation of the person deciphering them. Whether it's a film festival for humans, a friend of AI, the welfare of AI, or a signal from plants, the distance between them is determined by who and to what extent they confirm what AI is representing. Readers who receive the output of AI are required to be willing to check who has the people and mechanisms in place to verify what is being said on their behalf, before believing the output.

In this way, distance from AI is not just a matter of emotion, but also a choice about who is responsible and what to entrust funds and trust to.

What can be seen throughout each section is that AI investment decisions have shifted from ``how much to invest?'' to ``where to invest, with whom, and under what conditions.'' There have been calls to be selective about investments that support the economy, and pharmaceutical companies have reconsidered the size and partners of their partnerships. Companies are putting a price on talent, and organizations are re-determining the division of labor between people and AI. Countries and businesses have begun to draw the line on who can run AI and who can stop it. In creative and relational situations, the answer to whether or not to accept something depends on what the place treats as real. In any case, the size of the forehead is only the starting point for judgment. Readers are urged to check the intended purpose of the investment, its counterparties and carriers, applicable regulations and acceptable conditions before investing.

Q1: Who currently decides which parts of their work should be entrusted to AI and which parts should be handled by humans, and based on what criteria? Q2: Under what country's regulations and operator authority is the AI supplier you are using, and are there any procedures in place to accommodate changes? Q3: Next time you allocate funds to AI, other than the amount of equipment, which criteria will you consider in your decision-making, such as the partner, human resources, or purpose?

Finally, three questions

- Who is currently deciding which parts of their work should be left to AI and which parts should be handled by humans? Q2: Under what country's regulations and operator authority is the AI supplier you are using, and are there any procedures in place to accommodate changes? Q3: Next time you allocate funds to AI, other than the amount of equipment, which criteria will you consider in your decision-making, such as the partner, human resources, or purpose? - Under what country's regulations and operator authority is the AI supplier you are using, and are there procedures in place to accommodate changes? Q3: Next time you allocate funds to AI, other than the amount of equipment, which criteria will you consider in your decision-making, such as the partner, human resources, or purpose? - Next time when allocating funds to AI, apart from the amount of equipment, which criteria will you consider in your judgment, such as the partner, human resources, or purpose?