The Oura press release cited “that it is postponing its previously announced initial public offering on Nasdaq, despite strong demand, due to uncertainty in the IPO market”. No new date was given. Uncertainty is a factor, but there are others specific to Oura.

If you read Endpoints, they cited the Federal Reserve raising its rates a few weeks ago, depressing demand for the rich valuation Oura was seeking; the P(doom) around AI; and Apple introducing a new Watch with improved sensors and an option to order blood tests, as Oura has. There are rumors that Apple will also develop a faceless watch that will copy everything that Oura has. That’s a lot of putting off right there.

Ace investigator Sergei Polevikov in his AI Health Uncut of 29 September digs considerably deeper. Non-subscribers to his Substack platform will see some of this but his main points are that the oversubscribing level was weak at ~4x whereas ~20x for large offerings has been the norm since 2023. Three early VCs-Forerunner, Lifeline, and Elysian Park Ventures–were seeking to exit, never a good sign, and taking $1.53 billion with them. The IPO is also being built on Oura’s promise to deliver predictive health. Whether a ring–a piece of Finnish jewelry–can deliver that and more, reliably and interestingly to a growing user base, is a real question.

What 4x versus 20x says is that this IPO is a ‘nice to have’, not a ‘must have’. No one here has FOMO–fear of missing out.

There are hardware issues–and hardware is the vast bulk of its sales with service subscriptions only 19.8% of revenue.

Last but certainly not least, Oura’s benefit in going public on Nasdaq would largely have gone to the Taxman. Little known fact unless you’ve Been There, Done That: RSUs are treated as employee pay when they vest. The tax is withheld, just like wages. It becomes the employer responsibility, not the employee’s. In net settlement, Oura pays the tax in cash from its IPO proceeds and keeps part of the shares to cover the tax. (That doesn’t count what states do with RSUs nor how laws differ internationally.)

The line on the bottom for Oura after the IPO? The company nets only a paltry $6.2 million because of investor payoff and employee RSU taxation. That is over 98% of the proceeds going to everyone but Oura. 

Have other IPOs been like this? Not many. Evidently a lot of pay went to RSUs spread generously around, not just executive staff. This IPO evidently is to help three major VCs exit nicely–and a ~4x oversubscription wasn’t enough.

Another Ominous Parallel with an IPO that some of us remember: Peloton, once a fitness darling. There are similarities (the enthusiasm, the unfavorable hardware issues, the imbalance between hardware and subscription revenue, unsustainable growth beyond enthusiasts) and differences (the Oura Ring is under $400, a Peloton machine over $2,000, and Oura is profitable). But you’ll have to subscribe (and you should!!) to Sergei’s Substack to see all of his finely worked out and fully buttressed argument.

Oura is not exactly hurting for financing, having raised $1.5 billion over the past decade, with a jumbo $900 million Series E last October led by Fidelity, and an undisclosed corporate round this past July led by Eli Lilly, neither of which are exiting.  Crunchbase

Will Oura IPO this year or next? This Editor cannot see anything that was cited as changing the picture if the IPO were tomorrow or in the next few months.  Also Mobihealthnews

About the Anthropic IPO, the financial press is agog with talk that its engines may restart by mid-November. The date is rumored to be 9 November. Anthropic executives will meet with investors on 14 October to build a case for a $2 trillion offering, which would be an all-time record for an IPO.

Readers know that the IPO talk, explored since May, came to a screeching halt right around 9/11 with the talk of P(doom) [TTA 17 Sept] and Federal regulation. Some believe that these ‘external brakes’ were applied because neither Anthropic nor OpenAI were ready to IPO and it was easier to create Doom than to postpone directly. But here we have it at least for Anthropic, even if the fundamentals scream “warning”! Certainly the incentive to beat OpenAI to the IPO is still there. Anthropic, which was created by OpenAI bolters, surely want to ring that bell on either Nasdaq or NYSE (another small matter to finalize) before Sam Altman. Here we go again! Yahoo Finance

The prospective buyer, the Canadian branch of the Weston family, already is a giant in the Canadian food business with Loblaw, Real Canadian Superstore and other brands. They’re also experienced in the retail drug business with over a decade of owning Shoppers Drug Mart, a large national pharmacy chain. The family holding companies are Wittington Investments Limited and the public company George Weston Limited. Their last UK venture was Selfridges, which was sold in 2022 for £4 billion. Another branch of the family is an investor in Associated British Foods.

The Guardian helpfully adds that Boots has 1,800 stores across the UK and employs about 51,000 people, including about 6,000 at its headquarters in Beeston, three miles south-west of Nottingham. One wonders whether Boots will continue to sell Boots Beauty products such as No. 7 in Walgreens USA stores.

This summer, Sycamore tried and failed to sell Boots for $10 billion to Australia’s pharmacy group Sigma Healthcare. In London, a Weston buy has dashed hopes that Boots would be spun off and listed on the London Stock Exchange as reported in the FT. Those had multiplied after Boots in May appointed Alex Baldock, former boss of retailer Currys, as its new chief executive.

Walgreens bought Boots in two stages, with a 45% interest in 2012 and the remainder in 2014. The total price between cash and stock was in the vicinity of $15 to $16 billion. Yes, selling it for $9 billion along with associated debt is quite the ‘haircut’. 

Boots has been up for sale ever since the Sycamore Partners’ Walgreens Boots Alliance acquisition in March 2025 for a total value of $23.7 billion including debt, leases, and other factors [TTA 11 Mar 2025]. Sycamore took on an 83% debt level in doing so. Almost immediately, Sycamore split WBA into five parts, including Walgreens retail stores, Shields Health Solutions specialty pharmacy, CareCentrix, and VillageMD. Practices of the last have been either sold off in parts or shuttered, with Summit Health/CityMD remaining.

Yet, according to the FT, Boots is doing well. “The company reported in June that new beauty brands and the uptake of weight-loss jabs had driven up both its retail and pharmacy sales in the UK. Overall revenues rose by 3.2 per cent to £7.5bn in the year to the end of August 2025. Pre-tax profits jumped by a quarter to £337mn, driven by the reversal of impairment charges.”

What it means for Walgreens? Sycamore gets a quick cash infusion, allowing them to focus on revitalizing the US retail operation which had fallen on difficult times over the past three years. An indicator is a late September report that it slowed store closures from a projected 700 this year to less than 100, stabilizing total locations at about 8,000. Drug Store News Walgreens does not own a pharmacy benefit management (PBM) operation, a debit which now may be to its benefit as PBMs face financial and regulatory headwinds.

Both Sycamore and Weston were remaining mum about the deal to the press. Yahoo Finance UK, Axios

TTA has an open invitation to industry leaders to contribute to our Perspectives non-promotional opinion and thought leadership area. Today’s topic concerns youth mental health–and how specialized telehealth can provide remote psychiatric backup to assist pediatricians in making critical treatment decisions. The author, Andy Flanagan, is CEO of Iris Telehealth, where he draws on experience as a three-time CEO and senior leadership roles at Siemens Healthcare, SAP, and Xerox to guide the company’s mission of improving behavioral health outcomes for patients and clinicians.

Meta’s $16.7 billion settlement with a coalition of state attorneys general puts a number on something health systems have felt for years without a dollar figure attached. Youth mental health has become a public health infrastructure problem, and the infrastructure absorbing most of the impact isn’t behavioral health. It’s primary care.

New survey data from Iris Telehealth shows that when a child starts struggling with attention, mood, or behavior, 42% of parents try home strategies first. Another 23% bring the issue to a teacher or school counselor. Only 11% go directly to a mental health professional. Everyone else eventually lands somewhere else first, and for most families, that somewhere is a pediatrician’s office.

Pediatricians are absorbing psychiatric decisions that used to require a specialist referral, and telehealth can put that backup in the room with them instead.

Pediatricians have become the default specialist

Sixty percent of parents in our survey said their pediatrician or healthcare provider is where they get information on supporting their child’s behavioral or emotional health, ahead of friends and family, online resources, and school staff combined.

The same pediatrician parents go to for guidance is often the one prescribing the medication that guidance leads to. Among parents whose children have taken medication for a behavioral or emotional challenge, 44% said a pediatrician wrote it. Only 38% said a child psychiatrist did.

A prescription for a child’s mood or attention shapes how they sleep, eat, and function every day. Pediatricians are handling more of that decision themselves, and a child psychiatrist is rarely in the building to weigh in alongside them.

Fewer pediatricians will be available right when more families need one

The pediatricians absorbing this load today are about to become scarcer. A study published this year projects that pediatrician demand will grow 2% between 2025 and 2037, while the supply of pediatricians actually available to see patients drops by more than 10% over that same period.

Workforce adequacy, the measure of how well supply meets demand, falls from 92.3% today to 81.2% by 2037. Families in non-metropolitan areas and the South will feel that gap hardest, and general pediatrics is already projected to rank 14th out of 21 medical specialties for workforce adequacy by then.

At the same time, the number of families raising a mental health concern at a pediatric visit keeps climbing. A JAMA Network Open study of 1.8 million children’s insurance claims found that visits involving a mental health diagnosis rose from 5.7% in 2014 to 9.7% in 2023. Anxiety drove most of that increase, with related visits climbing more than 250% over the decade.

For a pediatric practice, that means fewer providers on staff and more children walking in with a mental health concern attached to their visit, a combination that leaves less time and specialized support for each one. Parents aren’t asking to skip straight to medication, either. Only 7% think it should be the first step, and 45% want skill-building tried before it. What they don’t have is a clear signal for when to move from that skill-building to professional guidance, the equivalent of a pediatrician tracking height and weight at every visit regardless of symptoms.

Integration bridges the pediatric access gap that primary care can’t close alone

Pediatricians shouldn’t have to become child psychiatrists. What they need is specialist support embedded directly into the visits already on their calendar. By integrating telebehavioral health clinicians into existing primary care workflows, health systems put expert guidance right at the point of care — specifically when specialists make critical prescribing decisions. The pediatrician is no longer left managing complex cases in isolation simply because the nearest child psychiatrist is half an hour away, or completely non-existent in their county.

AI plays a valuable, supporting role in this integrated model that’s strictly focused on efficiency. Used for triage support, ambient documentation, and identifying potential referrals, AI lightens the administrative burden that often makes specialist collaboration hard to sustain. It supports the clinical team, but it never makes clinical decisions.

While policy shifts, like Meta’s recent settlement over youth app safety, target the digital environment straining kids’ mental health, regulatory fixes don’t solve the immediate clinical access crisis. That requires accessible, expert care. Telebehavioral integration gives health systems the scalable infrastructure needed to support pediatricians today, starting with the very next patient on the schedule.

Oracle trumpets successes to Wall Street while continuing layoffs, tightening spending.

This perked interest by investors. Barclays has upped its Oracle price target to $252 from $250 and kept an “Overweight” rating on the tech stock. The stock price is currently around $154, down 54% from its high. The Street

This is despite that debt is high and building. Total debt climbed to $155.9 billion on a trailing twelve-month basis, up from $90.5 billion just two years ago. Net debt now sits at $118.9 billion.

More Oracle employees got their severance notes and cutoffs at the beginning of September. Restructuring has a price and it’s in people. Business Insider quoted ‘insiders’ that they started on 14 September with a chilly 6am note signed “Oracle Management” similar to the previous round: “After careful consideration of Oracle’s current business needs, we have made the decision to eliminate your role as part of a broader organizational change. As a result, today is your last working day.” Severance was 4 weeks’ base salary plus 1 week per year of employment. The count of the layoffs, LOBs, and states is unconfirmed by Oracle but estimated at 3,000, several hundred in Washington state. Not quite the ‘double digit percentages’ rumored in August but possibly so on some teams. This is on top of the earlier 21,000 global cuts, 13% of their workforce, originally posted as near 30,000 [TTA 31 March]. Oracle’s original restructuring cost estimate of $2.1 billion on severance payments and other costs linked to restructuring through 31 August was increased this quarter by over $700 million, bringing that total to $2.8 billion. NDTV, Quartz

Updated 22 Sept: It appears that the employee severance package for most people, in the tech context, is skimpy. It’s four weeks of base pay plus one additional week for each completed year of service, subject to a 26-week maximum. Microsoft’s maximum was up to 39 weeks. Moreover, Reddit employee threads state that the WARN period is deducted from the base pay (e.g. two weeks of WARN, two weeks of severance) and accrued vacation is lost if state law allows it. The article in TechTimes confirms that laid off employees forfeit unvested stock, but does not have how severance affects corporate bonus and the employee stock purchase plan.

Oracle’s new CFO Hilary Maxon denied in the next day’s all-hands that the layoffs did not mean that remaining employees would be doing more with less. CEO Mike Sicilia cheered the survivors on with “How does the work that I’m doing help deliver a better outcome for a customer?” Not an off-target ask, but if the work load does not change especially in healthcare or implementing an EHR, how does this situation not mean doing more with less? 2+2≠5

Rounding out a roller coaster two weeks for Oracle was a 19-0 House Veterans Affairs Committee subpoena for Larry Ellison and Mike Sicilia. It was voted on by the House Veterans Affairs Committee and issued before Labor Day, when Oracle did not attend the 2 September hearing citing scheduling conflicts. The hearing covered the $17 billion increase in budget for Oracle’s EHR development for current costs and the three-year extension to May 2031. It turned out to be rather raucous with accusations of “unreasonable” and “corruption.” The original $10 billion estimate was by Cerner and is running out [TTA 27 August]. The new budget request brings the total to $26.94 billion. Apparently the House members found out about it through news reports, While the House is not in session until 9 November, House committees can meet anytime, but both hearings are after the midterms. Sicilia is due on 19 November and Ellison on 10 December. FedScoop, Becker’s, Stars and Stripes, MedCity News, Healthcare IT News (updated for appearance dates)

Updated. Why Congress is Madder than Wet Hens is revealed in this FedScoop article and timeline. Simply and quickly, back in July 2022, when things started to go sideways in the EHRM implementation [TTA 28 July 2022], then-EVP for Industries Mike Sicilia told the Senate Veterans Affairs Committee that Oracle would absorb any “performance or workflow” issue costs above the original $10 billion ceiling. For Oracle, after the Transformational Big Vision kvelling faded, Cerner’s painful stumbles became Oracle’s VA Migraine. There are also other add-on costs for infrastructure related to the EHRM but not part of Oracle’s costs. 170% is a big surprise in the VA budget and it blindsided Congress.

TTA has an open invitation to industry leaders to contribute to our Perspectives non-promotional opinion and thought leadership area. Today’s topic concerns the disconnect between what gets funded–patient-facing tech–versus the more complex infrastructure technology that healthcare organizations need to “keep the lights on”

Digital health capital has a blind spot. It rewards what patients can see and touch, such as the app, the AI assistant, or the sleek intake screen, far more than it rewards the financial machinery that determines whether a company survives long enough to keep serving them. Carbon Health is the clearest recent illustration. In February 2026, the hybrid primary care company filed for Chapter 11 bankruptcy after raising more than $600 million over a decade and building roughly 90 clinics across eight states that together treated well over 800,000 patients a year. Nobody was complaining about the care. What forced the filing was a cost structure built for a larger company in a friendlier capital environment, and a balance sheet that no longer matched the business underneath it, according to court filings and reporting on the case. Patients experienced a functioning clinic right up until the day the financial layer underneath it gave out.

Carbon Health is not an outlier so much as the latest entry in a pattern that has repeated through several funding cycles now. Olive AI raised close to $900 million on the promise of automating revenue cycle work, then shut down in 2023 after its technology failed to deliver the savings it had promised hospitals. Pear Therapeutics and Babylon Health, two of the most recognized digital health brands of the pandemic era, went through bankruptcy the same year. The products differed. The failure mode did not: each company built something patients or providers could see and use, while the financial machinery underneath it stayed thin, brittle, or simply unbuilt.

This is worth naming plainly because the funding data suggests the pattern is still being set up, not corrected. Digital health investment is recovering in 2026, and AI is the reason why. U.S. digital health startups raised $7.4 billion across 244 deals in the first half of 2026, up from $6.4 billion a year earlier, according to Rock Health. Mental health platforms and GLP-1/weight-management startups remain the two most heavily funded clinical categories, and nearly 60 percent of all digital health investment in the first quarter came from a dozen mega-deals of $100 million or more, mostly consumer-facing. Capital is concentrating into fewer, larger bets, and visible keeps meaning patient-facing.

Meanwhile, the category that actually keeps healthcare companies solvent is losing ground with investors even as demand for it grows. Research from Galen Growth tracking healthcare buying behavior found that infrastructure’s share of health-system partnerships has climbed steadily, from under 19 percent of deals in the first half of 2022 to nearly 23 percent in the first half of 2026. Hospitals increasingly want the unglamorous layer underneath documentation, claims, scheduling, and care coordination, not another point solution. Yet over the same stretch, infrastructure’s share of venture financing dollars actually fell, and the number of infrastructure deals dropped by roughly 62 percent even as average deal size for the survivors doubled. Buyer demand for durable financial plumbing is rising, and the number of companies funded to build it is shrinking. That is not an absence of demand. It is a mismatch between what health systems need and what investors will write checks for.

The gap shows up most clearly in revenue cycle work, which is where a healthcare company’s survival is actually decided. Experian Health’s State of Claims research found that 41 percent of providers now report denial rates above 10 percent, up from 30 percent in 2022, and hospitals collectively spend close to $20 billion a year just overturning claims that were denied incorrectly. None of that shows up in a product demo. A denial queue does not make for a good conference stage moment the way a patient-facing app or an AI scribe does. It just quietly determines, months later, whether a company can make payroll.

That asymmetry is not really about technology. It is about what is easy to show and easy to fund. A patient portal or an AI assistant can be demonstrated in a five-minute pitch and photographed for a press release. A well-run eligibility check or a denial-prevention workflow is judged by what does not happen: the claim that does not bounce, the appeal that never has to be filed. Investors and journalists are simply better equipped to evaluate the former than the latter, and capital tends to follow what can be judged quickly. The cost of that bias never stays confined to the layer that was underfunded. It falls on the whole company, patient-facing product included, the way it did at Carbon Health.

None of this argues for less investment in patient experience or clinical AI; both remain genuinely important, and the returning capital in 2026 reflects real progress on that front. It argues for a more honest accounting of what digital health actually needs to survive its own growth. Until the layer that keeps the lights on gets judged by the same standard as the layer patients see, the industry should expect more companies that patients loved right up until the day they were gone.

A longish holiday does provide some Perspective of the Gimlety View.

Axios, the techie bulletin board and for generalist media types a source for pithy quotes, on 9 September quoted a MIT 2025 study of “experts” giving a 21.5% chance of “AI possessing dangerous capabilities”, with a 21% chance of AI initiating “Cyberattacks, weapon development or use, and mass harm”.  It was a reasonably analytic article full of nuggets for a writer to run with.

This past week introduced us to the latest pop tech slang, P(doom), a cute term for the percent chance of AI Doomsday. P(doom) has now surfaced everywhere from Elon Musk to Anthropic’s CEO Dario Amodei. (Impress your friends and relatives at the next cocktail party or tailgating/cookout.)

The mass media in the following week ran with quotes by Coxon, old tech stalwarts, and politicians. Plenty of P(doom) predictions. Emblematic of the coverage was the Los Angeles Times recap published on, ironically, 11 September, a day of real, not theoretical, doom.

Readers know from my running coverage that this Editor is skeptical of AI both in a business sense (a financial time bomb) and of its utility in its current models. Automating many functions within healthcare? Sure. Providing deeper insights into imaging to speed diagnosis? Faster drug development? Bravo. Speeding market and marketing analysis, sussing out needs? Hey, hey! But equally I have been scathing of companies that buy AI services and believe vendor promises, without cross-checking for accuracy, dumping the skilled humans doing these jobs in mass layoffs. These people are the boots on the ground who should be doing the cross-checks and upgrading the models that come from developers usually unfamiliar with medicine and healthcare–and certainly without deep and specific knowledge the boots have. It turns out that the mass layoffs in many companies, from team managers, data analytics, customer services, marketers to operations, have been to fund so-called investments and token spending in AI tools. ROI has gone into a Missing Man Formation in the finance department, while services to customers and in healthcare, patients, go sideways–as does market share and revenue. CFOs, funders and Mr. Market are just waking up to this surprise that they forgot to teach at HBS. 

I’ve also been scathing in coverage of certain companies such as Open AI, Oracle, and somewhat so of Anthropic, Microsoft, Meta, and Google. These companies are pushing one to two year bricks-and-mortar data center development without responsibility for local effects on water, power, noise, and land use–something that a GM, Ford, Toyota, or Worldwide Widgets building a factory would face and mitigate. This thoughtless pushing is proving to be politically evenhanded in opposition; these companies have received epic wedgies and a few pitchforks in their metaphorical derrieres as a deserved result. But this is manageable; because of pushback, it’s diminishing as a ‘reason why’ except as a political club for unscrupulous politicians.

What’s a lot more devastating is that apparently all the forecasting is inflated beyond belief. There is no decent idea of metrics, of matching up capacity to future demand. Not by OpenAI, not by Anthropic as they prep for IPOs. This is an ancient problem that always, reliably, cyclically bites developing sectors in their nether regions. Not a hype cycle like we saw circa 2006-15 with telehealth, not bad management, but in other industries within recent memory, with ‘sky’s the limit’ booms followed by a crash and partial/total hull loss:

TNW discusses in a deeper dive the debt structure and why PIMCO could make this bet where banks could not [reference is to Oracle’s debt funding]. The question it raises is whether the furious pace of data center building is another cycle of overbuilding–and if it is, will it be absorbed in time? The ominous parallels: the 2000s building boom in an earlier iteration of data centers, the fiberoptic boom of the early 2000s that broke WorldCom, Global Crossing, Winstar, Corning, and 360Networks, cloud overbuilding that left Amazon Web Services with years of excess capacity (it helps to have a deep-pocketed and not all that transparent parent), and others. This Editor would also liken it to the early years of 1980s-90s airline deregulation (too many airlines, too much debt, too many seats) and about a decade in the cruise ship industry where too many cabins were chasing too few people. These took decades and multiple bankruptcies to settle. TTA 7 May

Writer Ed Zitron, who has few parallels in Gimlety-ness, analyzed how OpenAI could easily meet the Devil of Demise sooner rather than later in his article What Happens If OpenAI Dies?, one of our Must Reads of 19 August. OpenAI’s losses are terrifying, contrary to their PR spin, and will not change in the immediate future; based on the numbers, their growth is slowing. It has to become the most successful company since Caesar Crossed The Rubicon–or it croaks and may well pull down an economy with it. I will add from other reading that Anthropic, ahead of OpenAI, is painted in a similar corner.

Leaving aside P(doom), what we are hearing are hoofbeats from horses, not zebras. What’s happening is the AI boom cooling and billions in concentrated funding that has hit multiple headwinds. Both companies have trillion-dollar valuations and heavy debt. Technical and security concerns are popping up like weeds (see the Hugging Face sandbox breakout).  Failures might not be the end of the world but not a cheerful earful for a stressed economy.

In the short term, the two major AI companies have more of a challenge from F(doom)–financial doom–than P(doom).

Suddenly, we have three things happening in the past 10 days:

• A short-term, low-level former Anthropic employee, Jacob Coxon, with few followers on X, suddenly bursting out and ‘going wide’ with a doomsday prediction. Quoted and covered everywhere in mass media with the things that at least some of his elders like Steve Wozniak and Geoffrey Hinton have been saying for over three years. (The Woz I’ve traced back to 2015!)
• Dario Amodei, CEO of Anthropic, Sam Altman, still CEO of OpenAI, and Elon Musk (Grok/X) then calling for government safety regulation. Not guardrails. Government regulation by a Federal agency, NIST. CNBC, OpenAI blog, Amodei blog statement
• Online ‘surveys’ stating that the public is now terrified of AI, with 63% rating Humanity’s Destruction from ‘moderate risk’ to ‘almost certain to happen’. (POLITICO) 

Private companies demanding government regulation of their business truly boggles the mind. This should raise red flags as to why, and why now, and who writes those regulations.

P(doom) supposedly has a ten-year horizon. F(doom) works far faster, may be more likely, and once it starts, there may not be an Anthropic or OpenAI to worry about.

Some Gimlety Speculation, IMO only:

• Both Anthropic and OpenAI float on a sea of red ink and Other People’s Money (OPM). Their lenders want to monetize and exit their investment with an IPO. Both companies were racing to IPOs. Except neither company is logically, reasonably, in any financial sense, ready.  Yet Anthropic was due to IPO before end of 2026 with OpenAI racing to beat them.
• Investors are also pushing for another OpenAI funding round to cash out secondary shares held by employees, according to CNBC today (16 Sept). Which, in a case of an IPO, means more shares for the investors.
• Both the Coxon media pickup and the subsequent Amodei/Altman calls for regulation effectively stop the IPO clock. One can speculate about the timing and origins, but it is fortuitous timing for both companies. .

It is also being used as a political club in the upcoming midterms. Goosing the ‘terror’ narrative is calculated to pay off certain players.

• The usual Senate suspects are demanding complete regulation of AI because “we are losing control”. There is a bipartisan bill in the House (Lawler/Gottheimer) drafted to regulate frontier AI, kicking the standards over to the National Institute of Standards and Technology (NIST), but the House is now out of session until 9 November, after the midterms, which leaves exactly two months before a new House is seated. There is a parallel bill in the Senate.
• President Trump, somewhat on the back foot on this, calls their fears a ‘hoax’ and a ‘scam’ that will prevent US leadership in AI development. Yet Sam Altman along with Nvidia’s Jensen Huang will be attending the state dinner for Chinese President Xi Jinping’s visit to Washington next week,
• The security threat is real. China is building a lot of data centers, well away of course from any international observation; Uyghurs don’t get to complain about data centers and water usage to Beijing. One doesn’t even know if they are real or empty buildings like their empty hinterland cities. Where are the chips coming from? Yet Chinese-engineered AI models may take the lower end of the market, much like steel.

There’s an old saying in medicine: when you hear hoofbeats, think first of horses, not zebras. It dates back to Dr. Theodore Woodward of the University of Maryland medical school and the 1940s in teaching his students to first rule in or out common diseases rather than searching first for exotic, rare diseases because they’re more interesting. (Apologies to those with rare diseases, and may AI be a savior in this.) Taking the analogy to business, the ‘horses’ are the shaky business models around AI development, capacity versus demand, the heavy debt of the two AI leaders, and data centers. Throw in the Chinese capacity and models. The ‘zebras’ are Humanity’s Extinction By AI. They’re there, but in the next county.

What’s running in that herd out there, who’s trying to paint the horses black and white, and why? Cui bono? The consequences can be severe.

Updated 17 September: Gary Marcus, Substacker and BBC commentator, in his latest ‘Marcus on AI’ takes an even more Gimletly view than your Editor. He points out the stratagems behind Altman’s and Amodei’s calls for regulation. It’s all ‘trust us’ to write the regulations and guarantee safety. We shouldn’t. Meanwhile, certain Senators go over the top on Doom with the aim of terrifying voters. We shouldn’t. A Must Read. And if you are working with AI, worth your subscription.

Silicon Valley Bank (SVB), now part of First Citizens Bank, is back with a just-published roundup of 2026’s first half (H1) in healthcare investment in the US and EU.

“Haves and have-nots. Winners and losers. The top of the heap and everyone else. Pick your metaphor….”

For health tech, it was the best H1 since H1 2022, with $7.2 billion versus $9.1 billion in investment, a tidy increase versus 2025 H1’s $6.7 billion. The major difference is that the number of deals is collapsing, with 203 deals this H1. If you look at 2025’s year total of 598 deals, if this deal track maintains, the 2026 full year number may not break 500. Again, the deal value is disproportionately and increasingly skewed to Series C+ deals, far more so than the past two years. Series B deals contracted and there was another uptick in Series A fundings. (Page 12)

SVB attributes the new health tech prosperity to AI-based technologies: clinical workflow automation, administrative efficiency and data infrastructure. Payviders (remember them?) also had their overdue day powered by AI: Devoted Health in the US with Medicare Advantage, and Alan in Paris (EU and Canada only). In SVB’s terms, “The bar for founders has risen toward durable growth, differentiated technology and a credible path to scale, especially those leveraging AI to address meaningful healthcare challenges.”

Exits, though, are hard to suss. There weren’t many this half for health tech. Kaia Health (Sword) in January, and in provider operations Ostro (Veeva) and Diligent (Serve). Kaia was the only one with a positive multiple on its valuation. Zelis and Virta Health might IPO this year or 2027. Or never. (Page 17). 

Looking at all four sectors, while investment dollars are growing, the number of deals continues to trend downward:

• H1 2026 healthcare investment at $24.7 billion is slightly below H2 2025 at $25.8 billion. The absolute high water mark was $43 billion set at H1 2021, in the midst of the ‘Throw Money At Anything’ period. But the trend is still one of recovery. 
• The big difference is that the number of deals continues its downward trend, sharply. H1 2026 had only 618 deals versus 2025’s H1 801 and H2 703. H2 is not projected to pick up this slack, again.
• 75% of investment is in the US, with EU accounting for 25%; EU is slowly but surely increasing its share.
• The falling investment sector is Dx (diagnostics)/Tools. Device remains steady and constant. (Health tech Readers should look over at device, since here resides WHOOP and its $575 million Series G for Giant) skewing this sector.

Most concerning is that healthcare VC fundraising has hit new lows, and the money is concentrated in a few across the seed, early, and late stages. Only a few years ago, founders had a wide choice of VC funders with fat wallets. Not nowadays. H1 2026 funding for VCs was only $10 billion. Full year 2025 was $16 billion and 2024 $26 billion. Many large “name” funds ($100 billion plus) closed this year, such as Lux, Santé, Frist Cressey Ventures, and Kleiner Perkins. What’s left in health tech? General Catalyst, Andreesen Horowitz, Redesign, Menlo Ventures. It’s a monotonous roster. Occasionally, there’s a crossover. Especially at early stages, founders have lost leverage in finding a simpatico fund. They have to show that they have a credible path to growth and profitability. (Pages 7-8)

Unlike SVB’s last report for 2025, they did not break out a fifth sector cutting across all four, Longevity and Healthspan. 

SVB will also be changing their name during Q4 2026 to First Citizens Innovation Banking. Preview page for SVB’s Healthcare Industry Trends published 25 August.  Hat tip to Megan Scheffel, head of SVB’s healthcare sector.

Oracle’s Seema Verma rumored to be on a list for a new CEO position with a major lobbying group.

ATA Action was kind enough to flag for this Editor that the Drug Enforcement Administration (DEA) may finally be moving toward a new rule for telemedicine prescribing of controlled substances. Yesterday (25 August) the “Special Registrations for Telemedicine and Limited State Telemedicine Registrations” final rule was submitted to the Office of Management and Budget’s (OMB) Office of Information and Regulatory Affairs (OIRA) for review. While under review, the text is not public, but the OMB pending EO  is here with the limited text under the RIN: 1117-AB40   The text of the RIN includes that “DEA is currently reviewing the over 6,400 public comments submitted on the Special Registration for Telemedicine NPRM (Notice of Proposed Rulemaking) published on January 17, 2025.”meaning that they are still working with a year-old NPRM.  DEA has extended telemedicine prescribing rules several times, most recently at the end of 2025 to cover full year 2026. A final action to determine approval of the rule is due in November, according to the timetable.

ATA Action’s statement, per their email, is “ATA Action is actively engaged as the rule moves through OMB review, pressing the priorities we have consistently advanced: protecting patient access while maintaining safeguards against diversion — without unnecessary in-person mandates, duplicative state-by-state registration requirements, excessive reporting burdens, or implementation timelines that make participation impractical for legitimate clinicians and healthcare organizations.” They also cite geographic “red flags” that pharmacies face when a prescription results from a telehealth visit and state laws that are more restrictive than Federal law. 

Brooklyn’s Cityblock Health’s goes country with Homeward Health acquisition in an all-stock transaction. It is on the surface an odd match: an urban-focused value-based care provider that primarily serves complex needs and provides additional support for Medicaid and Medicare/Medicaid dual-eligible patients expanding into underserved rural markets. What it does promise Cityblock is growth from its present claimed 200,000 members in 10 states and annualized revenue of $2.2 billion, up 77% year-over-year. Homeward serves around 50,000 rurally-based Medicare Advantage (only) members and 5,000 providers in Michigan and Minnesota through a Michigan clinic, mobile clinics and telehealth. It partners with Blue Cross Blue Shield of Michigan and Aetna for Medicare Advantage. Further financial arrangements were not disclosed.

Homeward’s CEO Dr. Jenny Schneider will remain connected with Cityblock as a “trusted advisor” according to a Cityblock spokesperson and Homeward’s brand and operation will remain. According to her LinkedIn post announcing the acquisition, she previously served as a Cityblock board member. She and another Livongo alumnus, Amar Kendale, launched Homeward Health in 2022 to close healthcare gaps in rural America, accounting for over 61% of primary care shortage areas and a 23% higher mortality rate.

Both Medicaid and Medicare Advantage face challenges–funding cuts for Medicaid and with Medicare Advantage, payers reducing reimbursement rates based on rising costs and increased Federal scrutiny of higher costs, upcoding, overpayments, and bonus programs. 23% of Medicare Advantage enrollees are in a special needs plan (SNP) KFF

Cityblock touts its CORE—Care Orchestration and Resourcing Engine—as an AI-based operating system underpinning our outcomes-based care platform to use almost 10 years of member data to close care gaps for specific members, as well as AI tools “handling routine engagement and administrative work that have already freed up 44,599 hours of clinician time this year, including 42,148 calls handled by AI agents.” Homeward Health’s platform is also AI-native.

Cityblock simultaneously announced a Series E of $116 million led by General Catalyst for a total of close to $1 billion. General Catalyst is an investor in both companies. (Is this a return to the General Catalyst ‘portfolio condensation’ strategy that created Commure last year?–Ed.)

Interestingly, Homeward Health is organized as a public benefit corporation (PBC) and a B Corp, certified by B Lab as meeting their standards of social and environmental performance, accountability, and transparency. It’s doubtful that will carry over to Cityblock. MedCityNews, FierceHealthcare, Becker’s, Cityblock’s CEO Toyin Ajayi blog post 

Biggar, Scotland senior housing environmental sensor-based technology test saves thousands, but across Scotland could save millions. The test at the retirement home Langvout Court, part of Bield Housing & Care, was of environmental and activity sensors developed and implemented by Glasgow-based technology company Archangel and the Digital Health & Care Innovation Centre, funded by the UK Government’s Department for Science, Innovation and Technology through the Glasgow City Region 5G Smart and Connected Places Programme. The automated alerts were directed and coordinated with Bield staff. The sensors alerted the staff to a potential boiler breakdown during Storm Éowyn, high humidity levels and faulty extractor fans in buildings, and resident temperature and activity changes that could indicate changes in health. Annualized savings in the six-month test were estimated at £7,670 in heating costs and £2,825 in maintenance costs, along with improved tenant safety, reduced manual checks and stronger regulatory compliance, creating positive ROI in year one. FarrPoint, which independently evaluated the project, extrapolated that applying Langvout Court’s system across Scotland’s retirement housing developments could generate annual savings of £18.5 million. Archangel case study published in TechUK, release in Care Sector Hub

Background: Oracle’s current VA contract 36C10B18D5000, the Electronic Health Record Modernization Indefinite-Delivery/Indefinite-Quantity (IDIQ) contract with Oracle Health Government Services, the successor to Cerner Government Services, originally was structured with a 10 year base period without annual renewals. This started in May 2018. In early 2023, the contract was renegotiated. The base period was halved to five years with the remainder of the contract subject to annual renewals from May 2023 and expiring 10 years after the contract start, in May 2028. The Modification P00008 to the IDIQ contract adds another three one-year firm fixed-priced optional ordering periods, taking the duration of the contract to May 2031.

According to the released contract modification documents, VA projects that the original ceiling would be reached by Q1 in Federal FY 2027, which is between October and December 2026. The rationale is that “Due to unanticipated complexities in deploying this system, extensive site-specific customizations, as well as the other factors … VA has reached the current contract ceiling sooner than originally planned. VA anticipates using the remaining ceiling by the first quarter of fiscal year 2027 and thus requires an increase.” 

“Building upon the Trump Administration’s successful deployment of VA’s new electronic health record system at several VA facilities, VA and Oracle Health agreed earlier this month to renew their collaboration,” VA spokesperson Quinn Slaven previously said in an emailed statement to FedScoop.

As previously noted in our original article [TTA 12 Aug], in this Editor’s view, the three-year extension is a smart move on the VA EHRM team’s part.

• The obvious one is that the VA EHRM rollout requires another three years from 2028 to 2031 to fully cover all locations. Releasing the main single-source contractor three years prior to its finalization is not an intelligent move unless another EHR company could take it up–and that has not happened. Things apparently are going smoothly in the newly formatted rollout. It was also obvious that the cost ceiling had to be increased.
• The other is protection. Now you won’t read this elsewhere. Since the late winter, Oracle was rumored to be interested in selling, wholly or in part, Oracle Health AI (OHAI). Oracle Health sale rumors were confirmed this summer. In the event of a sale, the buyer would be obligated to honor the VA contracts and its terms.
• The other possibility is if disaster strikes Oracle as a result of their AI landlord strategy, such as bankruptcy, the VA has some contractual protection in a Federal court. 

Orange Slices, NextGov/FCW

VA Indiana’s three-hospital system (Fort Wayne, Marion, and Indianapolis, VISN 3) cut over to the Oracle EHR on 22 August. X announcement (@VAIndyHealth)  The last two scheduled for 2026 will be Anchorage, Alaska (VISN 5) and Cleveland, Ohio (VISN 3) on 24 October.

Grab a cuppa and sit down with these articles. (You may also want to subscribe to their authors.)

From Substack, UK author Martyn Eeles current Health VC newsletter, “The Strategic Clarity Problem”, advises founders of early-stage companies that doing more can result in less–financing. Too much activity in too many directions leads to confusion on investors’ parts. Paradoxically, it doesn’t enhance “potential” but detracts. It reads to investors, especially now, as lack of priorities and not strategic. Mr. Eeles recommends focus, focus, focus. Choose a strategy and stick with it. It doesn’t mean that a founder cannot show multiple future paths, just that the main path has to carry the company forward. (Sounds like good marketing!)

For instance, how you present your direction is vital in making activity sound focused and strategic. FTA:

A founder who says, “There are many use cases,” may sound ambitious. A founder who says, “There are many possible use cases, but this one is the wedge because it creates the clearest buyer urgency,” sounds more investable.

A founder who says, “We have lots of partnership conversations,” may sound active. A founder who says, “These two partnerships matter because they reduce implementation risk and create access to the customer segment we are prioritising,” sounds strategic.

The short (non-subscriber) version has a wealth of information for both founders and funders, complete with a nifty infographic that depicts nearly the entire article. but truncates at ‘The Choices Investors Want To See”.  This Editor would recommend the annual €60 subscription if you’re in the business. Mr. Eeles is managing partner at Clarma Capital, a European life sciences venture fund.

Our friend Sergei Polevikov writing in his Substack AI Health Uncut returns to the General Catalyst-powered Commure in Commure’s Long History of Red Flags. Even though General Catalyst doesn’t want him to.  Yes, the flags still flap around the gaggle of health tech companies financed by General Catalyst (GC). Commure itself is an agglomeration of GC companies: Athelas, Augmedix, RxHealth, and Memora Health. Commure originally had one marketable product, Strongline, a safety and duress badging/tracking system, three years ago before GC’s consolidation moves. What is questionable about Commure has now surfaced in STAT News +’ investigation (paywalled). From pricing dependent on recommendations to products that don’t work until they’re modified at the client if they eventually do work, to referral programs that are way too close to violating the Anti-Kickback Statute…Commure has it all on the Shady Side of the Street. GC keeps shoveling money in because they can, too. 

And once again, thinking the unthinkable, is Ed Zitron. Here he imagines the demise of OpenAI and reads the tea leaves. He notes:

• the deceleration of revenue when it needs to accelerate (see below)
• the COO and CRO left after less than a year on the job, likely walking away from generous stock options/awards–now, who does this?
• it’s backed away from its IPO and likely will be beaten to it by Anthropic (Claude)
• the economics are terrifying. OpenAI lost $20.9 billion in 2025 on $13.07 billion in revenue
• it needs to meet compute obligations and for that needs $800 billion in cash
• it needs to raise $100-200 billion annually just to survive

In short, it has to become the most successful company since Caesar Crossed The Rubicon–or it croaks. Expires. Meets the Devil of Demise and the Devil wins.

The consequences will be severe. FTA:

To not actively and meaningfully discuss the potential for OpenAI to collapse is actively irresponsible. To act like there are not significant, existential problems with this company’s economics is to intentionally avoid reality, and whoever is on the receiving end of said ignorance deserves better, be they an investor reading your analyst note or a reader burdened with incomplete journalism.

What follows may be an Enron-Lehman Brothers hybrid, one that leaves unbelievable destruction in its wake, an avoidable systemic risk empowered and enabled by a kneecapped media industry and sell-side analysts incapable of seeing further than two quarters in the future.

The time to stop this? Long past.

Zitron backs everything up with hard numbers laced with cross-references. It’s dense and needs close attention. Depending on your view,  you’ll choose a gallon of coffee, a fifth of bourbon, or a bottle of wine. What Happens If OpenAI Dies?

Another argument, shorter, and similar, is made by Gary Marcus in his Substack newsletter (free access), Marcus on AI,  BREAKING: OpenAI’s unraveling has begun.  Again, just as it was scheduled for its IPO and racing its main competitor.

It contains two citations from the Wall Street Journal writers who cover OpenAI, Berber Jin and Corrie Dribusch:

• “The company grew revenue by just 18% to $6.7 billion from q1 to q2, while its losses sank further into the red”
• Losses grew from Q1 to Q2 to $3 billion to $12.3 billion, while it added only $1 billion (to $6.7 billion)

Nvidia is in full CYA mode, given its exposure to OpenAI. Can Oracle be far behind?